State of the art report: Analysis and design

Analyse et conception : un état de l’art

B. Simpson & P. Morrison

Arup Geotechnics, United Kingdom
S. Yasuda
Tokyo Denki University, Japan
B. Townsend
Hatch Mott MacDonald, LLC, USA
G. Gazetas
National Technical University of Athens, Greece

ABSTRACT
This paper is concerned with analysis and design across the breadth of geotechnical engineering. Sections by individual experts relate
to Codes and standards, Deep foundations, Embankments and slopes, Underground construction and Seismic design. Each considers
recent developments, including economy and sustainability. Recent geotechnical codes and standards from Europe, North America
and Japan are compared. All adopt forms of limit state design and address familiar geotechnical problems of uncertain ground
behaviour and complex interaction of loads with frictional materials. Some topics which are still under debate are discussed. Design
of deep foundations is a blend of empiricism and theory. Developments including drilled and grouted piles in rock and offshore
driven piles are presented. Topics considered include assessment of ultimate capacity and the environmental benefits of pile re-use
and piles as heat transfer elements. Design of embankments and slopes depends on empirical rules and observation, as well as theory.
In Japan, tolerance to displacements following earthquakes is a particular consideration. Embankments used as transport corridors or
as building platforms are discussed. Failures involving uncontrolled fills are particularly noted. For construction underground,
understanding of the fundamental physics is imperative, along with comprehensive process of checking during both design and
construction. Greater use of cement replacements would aid sustainability. The final section of the paper considers performance-
based design of foundations against two seismic hazards: emergence of a rupture underneath a structure, and bearing capacity
mechanisms for slender structures on shallow foundations.
RÉSUMÉ
L’article étudie l’analyse et la conception d’ouvrages géotechniques variés. Les différentes sections abordent les codes et standards,
les fondations profondes, les pentes et talus, les constructions souterraines, puis le génie parasismique. Les développements récents,
notamment en matière d’économie et de développement durable, sont présentés. Les codes et standards utilisés actuellement en
Europe, aux Etats-Unis, et au Japon sont comparés. Tous adoptent des formes de conception aux états limites et abordent les
problèmes courants causés par les incertitudes sur le comportement du sol, et par les interactions structures-milieux frictionnels.
Certains sujets encore en débat sont abordés. La conception de fondations profondes mélange empirisme et théorie. Plusieurs
développements sont présentés, tels les pieux forés-cimentés dans les roches et les pieux battus en mer. Estimation de la capacité
portante ultime, intérêt environnemental de la réutilisation de pieux, et utilisation des pieux pour transférer la chaleur sont, entre
autres, considérés. La conception des pentes et talus repose sur des lois empiriques, l’observation, et la théorie. Au Japon, la tolérance
aux déplacements dus aux séismes est d’une importance particulière. L’utilisation des talus comme couloirs de transport, ou comme
plateformes pour les constructions, est abordée. Les ruptures de remblais non contrôlés sont aussi étudiées. Les constructions
souterraines exigent une bonne compréhension de la physique fondamentale et la vérification des phases de conception et de
construction. Une utilisation plus fréquente de substituts au ciment serait plus environnementale. La dernière section concerne la
conception des fondations contre deux phénomènes liés aux séismes: apparition de discontinuités sous la structure, et modes de
ruptures par perte de capacité portante des fondations superficielles de constructions élancées.
Keywords design, analysis, codes, standards, piling, tunnelling, slopes, embankments, seismic

1 INTRODUCTION the design process. Analysis is one part of design, involving

The president of the ISSMGE, Professor Pedro Seco e Pinto,
invited an international group of experts to prepare this State of
the Art Report on Analysis and Design in geotechnical
engineering. It was agreed that five topics would be included in
the report: Codes and standards, Deep foundations, Slopes and
embankments, Underground construction and Seismic design.
Each author was asked to provide a brief review of the
fundamentals of technical understanding of most importance to
their topic and to comment in particular on recent
developments, especially since previous state-of-the-art papers
on the topic. Comments are provided on what goes wrong, and
emphasis is given to questions of economy and sustainability.
Design is decision. In the context of this paper, design is the
complete process or sequence of decisions which determine
what is actually built. The decisions may be taken by clients,
consultants, contractors, site workers or others – all are part of
calculations carried out by engineers, with or without the aid of
computers. The main thrust of the contributions relates to
design aspects. References are made to supporting analysis, and
it is assumed that advanced techniques such as finite element
computations will be available and used when appropriate.
2 CODES AND STANDARDS
2.1 Introduction
Codes and standards aim to set out the process and procedures
of design, or at least to identify the basic ingredients and limits
of acceptable good practice. Figure 2.1 shows the functions
performed by codes in many projects, providing the
communication between society, clients, data and analysis in the
process of design.

Proceedings of the 17th International Conference on Soil Mechanics and Geotechnical Engineering 2873
M. Hamza et al. (Eds.)
© 2009 IOS Press.
doi:10.3233/978-1-60750-031-5-2873
2874 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 2.1: Communication facilitated by codes
In this paper, the term “code” will be used in a broad sense.
It includes “design standards” such as the Eurocodes, “design
codes” such as Japanese codes, and “design specifications” such
as the AASHTO LRDF Bridge Design Specifications. The
older style of British “code of practice”, which often included a
more discursive account of the geotechnical design process
rather than definitive rules, is within the scope of this
discussion, but specifications for construction are outside the
scope.
Part of the function of a code is to provide a link between
analysis and design, showing how analysis can be used as a tool
in the design process. Codes may provide limits on the analysis
methods to be used, deliberate or unintended, and generally
provide safety factors or margins, establishing gaps between
what is analysed, failure states and what is most likely to occur
in practice.
Society requires safety and serviceability. Codes provide a
link between the activities of the designer and the requirements
of society. They are the point at which society’s qualitative
requirements that structures should be at least as reliable as they
traditionally have been are interpreted in the forms of
procedures and numbers. Society, or at least clients, also
requires economy, to be balanced with safety and serviceability.
Although codes have often promoted safety and serviceability
with less regard for economy, they also have a responsibility
here, especially in a world in which limitations of resources are
increasingly recognised.
Perhaps the greatest challenge for drafters of geotechnical
codes is to find the best balance between fixed rules and
personal expertise, both of which are essential to successful
design. Geotechnical engineering relies heavily on the
knowledge and judgement of individual designers. Inevitably
this is somewhat subjective, depending on the training and
experience of each individual, but equally it is indispensably
valuable, especially when shared in discussion with others;
codes must allow and encourage this. For both individuals and
the profession as a whole, past mistakes are a major source of
learning, and both must aim to avoid repetition of these. Codes
are an important part of the profession’s corporate memory,
helping to assure society that past mistakes will not be repeated.
Equally, though, modern codes are aiming to pull professional
practice forward, closer to the best available “state of the art”.
In the following sections, approaches currently taken by
codes to geotechnical analysis and design will be reviewed and
compared. Particular emphasis will be given to European,
North American and Japanese developments, noting both the
broad areas of agreement and particular points of debate within
and between these communities. Specifically, the main
documents included in the review are the Eurocodes,
particularly Eurocode 7 – EN1997-1 (2004), Geotechnical
design; the AASHTO LRFD Bridge Design Specifications
(2008); and Japanese Geocode 21: Principles for Foundation
Designs Grounded on a Performance-based Design Concept
(Japanese Geotechnical Society 2006).
The Eurocodes use the term “actions” in place of the term
“loads” used in American and older British codes. English
translations of Japanese documents use both terms. In this
paper, the term “loads” will generally be adopted, except in
direct quotations.
Clause numbers from codes are shown in brackets thus: { }.
2.2 Limit states and working states
Investigators of failures frequently report that the main cause
was some onerous condition which had not been considered in
the design. In contrast, it is often noted that failures were not
caused by normal or even somewhat excessive statistical
variations of parameters which had been properly reviewed and
for which standard margins of safety had been allowed. It
appears to be important, therefore, that designers’ attention is
drawn towards a reasonable range of extreme situations against
which to check each design. This is the underlying philosophy
of limit state design.
Limit states are “states beyond which the structure no longer
satisfies the relevant design criteria”. That is, the structure, or
part of it, is on the point of “going wrong” – failing in a sense
that would be understood by a client, user or non-technical
observer. All the codes place the limit state on the safe side of
the failure, ie when the structure is almost failing but has not yet
failed. Limit states are classified into various types, used
selectively in the codes.
In the Eurocodes, Ultimate limit states (ULS) are “associated
with collapse, or with other similar forms of structural failure”.
They generally relate to danger or severe economic loss.
Serviceability limit states (SLS) “correspond to conditions
beyond which specified service requirements for a structure or
structural member are no longer met”. These are unwanted but
more manageable if they occur, allowing the structure to remain
in service if necessary repairs are carried out. They may be
subdivided into irreversible SLS (eg cracking) where the
consequences of exceedance remain after the disturbing action
is removed and reversible SLS (eg unpleasant vibration) where
the consequences do not remain. The Eurocodes also consider
accidental design situations, though these are regarded as
potential causes of limit states, rather than limit states in
themselves.
The Japanese Geocode 21 emphasises that the ULS is still on
the safe side of collapse: “the structure may have sustained
considerable damage, but not to the extent that the structure has
reached failure, become unstable, collapsed, or to the extent that
would result in serious injury or the loss of life.” Similarly, at
SLS there is to be no effect on durability and “regular use of the
structure is possible, without repairs.” This code also has a
Reparability limit state “in which damage to the structure has
occurred and may have affected the durability of the structure.
However, regular use of the structure is possible to a limited
extent and there are reasonable prospects for full functionality
of the structure if economically-feasible repairs are performed.
This limit state can be interpreted as the state in which the
majority of the value of the structure has been preserved. In
addition, the reparability limit state sometimes implies a state in
which marginal use of the structure is possible for rescue
operations immediately following an extraordinary event such
as a large earthquake.”
The AASHTO Bridge Code has Strength limit states
“relating to strength and stability during the design life”. This
is a more technical definition than the Eurocodes’ ULS, but
from most points of view it serves a similar purpose. The same
code has Service limit states “relating to stress, deformation,
and cracking under regular operating conditions”. AASHTO
also have Extreme event limit states “relating to events such as
earthquakes, ice load, and vehicle and vessel collision, with
return periods in excess of the design life of the bridge.” These
 B. Simpson et al. / State of the art Report: Analysis and Design
may be compared with the Eurocodes’ accidental design
situations noted above.
Offshore codes such as Det Norske Veritas (1993) include
other “limit states” such as progressive collapse limit state,
fatigue limit state, though these may be considered more
descriptive of the way the limit state is reached than of the limit
state itself. The survivability limit state, however, generally
does describe a particular state of the structure.
In Limit State Design, attention is directed to states at or
close to failure, which it is hoped will not occur in practice.
The alternative is Working State Design in which attention is
directed to the expected, desired state in which the structure is
performing successfully under its expected loading. Its safety is
then usually checked by requiring that the degree to which
material strengths or ground resistances are mobilised is limited.
Limit state design codes usually require that at least two
limit states be checked. One of them is normally a
serviceability limit state, for which unfactored parameter values
are usually used in calculations. This is the nearest approach to
an analysis of the expected performance of the structure, but it
is more pessimistic than is really expected, as discussed further
below. The other limit states involve more severe situations,
often derived by applying partial factors to selected parameters
representing loads, ground or structural material strengths,
ground or structural resistances, and possibly other parameters
including water pressure.
Despite some differences of emphasis, or of drafting, the
developers of the modern codes agree that a limit state
approach, concentrating on failure, or near failure, is to be
preferred (see, for example, Becker, 1996). The present authors
agree with this conclusion, adding the following observations:
1. Limit state design implies that the states studied should not
occur. For ULS, their probability is to be extremely low, so
calculations are directed to a virtually non-existent state.
This may confuse designers who more readily imagine
states which they consider fairly likely to exist. In this
respect, working state design is easier to understand.
2. It was noted above that actual failures are often caused by
the occurrence of situations that had not been properly
foreseen during the design process. Some of these might be
accepted as virtually unforeseeable, while others are errors
of process on the part of the designer, such as misjudgement
of soil properties or failure to check a critical mechanism. It
is essential, therefore, that codes continue to emphasise
good geotechnical practice and process, avoiding too great a
concentration or reliance on factors of safety.
3. Calculation errors are discovered very often during
investigations of failures. Complexity in codes seems likely
to increase the number of errors, so it is important, as a
safety issue, that code provisions are kept as simple as
possible.
4. A few of the calculation errors prove to be critical, but it
seems likely that many others are covered by the margins of
safety traditionally in use. Human error is therefore a
significant uncertainty that must be included in any rational
attempt to determine factors of safety. Failing this, overall
factors of safety can only be reduced with the utmost
caution.
5. The aim, however, should be to reduce factors of safety, and
certainly not to increase them, in general. Unnecessarily
large factors of safety lead to increase in cost, and wastage
of materials and energy.
2.3 Assessment of parameter values – characteristic values
Prescribed values for partial factors provided by codes have
little meaning unless it is clear how the parameter values are
initially to be selected before being factored. These unfactored
parameters are referred to as characteristic values of loads,
strengths or resistances in Eurocodes and Geocode 21, and as
nominal values in AASHTO. In effect, modern codes provide
2875
safety margins by a combination of two features: somewhat
cautious characteristic values, determined by the designer in the
case of soil properties, and the application of partial factors,
prescribed by the codes on behalf of society. Usually,
“cautious” values of strength are lower than the most probable,
but for some limit states values on the high side are more
critical.
In the case of loading, the values themselves are largely
prescribed by the loading codes. For manufactured materials,
characteristic values of strength are derived as a fractile of the
test results of a closely specified type of test, typically 2
standard deviations from the mean, giving a frequency of about
2.3% in the tests. It is commonly the case that code drafters are
more knowledgeable about both loading and structural material
properties than are designers. For ground material the situation
is more complex, however, for several reasons: (a) ground
materials cover a wide range of types, some more easily
sampled than others, so a range of testing approaches is
unavoidably adopted; (b) in most, but not all cases, the
parameter relevant to the design is close to a mean value,
averaged over a large zone or surface within the ground, rather
than to an extreme derived from testing small samples; (c) the
uncertainty and variability of the ground on a particular site is
often much better known to the designer than it could be to the
code drafter; (d) the process of construction sometimes changes
the ground properties; (e) data and observations obtained from
similar projects may be highly relevant to selection of
appropriate parameter values; (f) full scale testing is quite often
incorporated into the design process, particularly for piles and
ground anchors.
This problem is not new: in past practice it was usually
unclear how conservative the code drafters assumed the values
of parameters to be. An early attempt at providing more
guidance was represented by CIRIA Report 104 (Padfield and
Mair 1984) which asked users to assess either “moderately
conservative” or “worst credible” values for soil strength
parameters and provided differing factors for these two cases.
The same approach is taken in the more recent replacement of
this report, CIRIA Report C580 (Gaba et al 2003).
The developers of various geotechnical codes have taken
different approaches to this problem. After very lengthy
debates in Eurocode 7, the basic definition of characteristic soil
parameters was agreed as: “a cautious estimate of the value
affecting the occurrence of the limit state”. Additional text
makes it clear that this estimate is to include allowance for the
extent of ground involved in the limit state, and hence average
effects over that extent, and the effect of construction activities
on ground properties. In making this estimate, input from
laboratory and field tests, and also well-established experience
are required, but the type of testing is not restricted and a
distinction between situations in which spatially averaged
values or extreme values govern is encouraged. It is suggested
that if statistical approaches are applied to the data “the
characteristic value should be derived such that the calculated
probability of a worse value governing the occurrence of the
limit state under consideration is not greater than 5%”. Thus the
emphasis throughout is on engineering assessment of the values
actually governing behaviour in the ground, which might differ
from those measured in tests, with a requirement that the
assessment be “cautious”, not a best estimate of the most likely
value, especially if there is considerable uncertainty. It is
implied that “cautious” means that the chance that values worse
than the selected value will actually occur in such a way as to be
the cause of a real problem (a limit state) is about 5%. The
expertise of the engineer is by no means relegated, but is
incorporated into the assessment of the characteristic values.
The present authors would argue that this is no different from
previous good practice, and essentially the same as the
“moderately conservative” values of CIRIA Report 104;
engineers have generally aimed for a mean value, but have been
2876 B. Simpson et al. / State of the art Report: Analysis and Design
slightly cautious, especially when the degree of uncertainty was
significant.
It is important to recognise that in a typical situation
involving large amounts of ground a “probability of a worse
value governing the occurrence of the limit state under
consideration is not greater than 5%” is very different from a
5% fractile of test results. Schneider (1997) discussed situations
where the zone of ground affected is large enough that its
behaviour will be close to the overall average of the soil
considered. He argued that a “cautious estimate” with a 5%
chance of a worse overall value would lie about 0.5 standard
deviations from the mean of relevant measured values, adjusted
as necessary for sampling disturbance and construction effects.
The marked difference between these two assessments can be
seen in Figure 2.2.
Figure 2.2: Alternative derivations of “characteristic values”
A similar proposal was made by Dahlberg and Ronold
(1993) for design of offshore foundations and recommended by
Becker (1996) for more general use. This involves the use of a
“conservatively assessed mean” (CAM) as the characteristic
value, also about 0.5 standard deviations from the mean of the
test results. These authors state that for a normal distribution
75% of the measured values would be expected to exceed this
value. (More accurately, this requires an offset of 0.69 standard
deviations from the mean, for a normal distribution, as shown in
Figure 2.2). Foye et al (2006b) take up the same idea proposing
to use a CAM with 80% exceedance, equivalent to 0.84
standard deviations below the mean of the test results for a
normal distribution. These proposals are made in the
development of North American practice, though at present the
AASHTO Specifications do not use the term “characteristic”
but refer less specifically to nominal values related to
“permissible stresses, deformations, or specified strength of
materials”.
Japanese Geocode 21 {2.4.3} defines characteristic values of
geotechnical parameters as “the representative values that have
been cautiously estimated as the most appropriate values for the
foundation-ground models in order to predict the limit states
checked in the design”; this is apparently close to the Eurocode
definition. It continues: “The characteristic value of a
geotechnical parameter is principally thought of as the average
(expected value) of the derived values”, but it adds
qualifications to this statement that diverge away from a simple
statistical average, and shows the same in a mathematical
formula. The terms “cautiously estimated” and “expected” are
potentially in conflict, but the overall impression given is that
the characteristic value may be somewhat more conservative
than the statistical mean. However, commenting on Geocode
21, Honjo et al (2005) particularly emphasise that “the
characteristic value is defined as a mean value of a geotechnical
parameter”.
Thus there seems to be approximate agreement between the
Eurocodes and the North American proposals about the level of
conservatism most usefully associated with the characteristic
value; the Japanese position might be slightly different. The
recommendations of Becker and of Foye et al, intended for
North American practice, encourage the adoption of a limited
range of specific tests, CPT and SPT, with the characteristic
value derived directly from the test results (including
modifications for stress level and possible removal of rogue
values). This would largely remove judgement from the
process, although considerable geological judgement remains in
zoning the site, both in plan and in level. In contrast, the
European approach incorporates the designer’s expertise into
the assessment of characteristic values. It allows assimilation of
results from a range of test types and sources, and encourages
engineers to consider the relevance of the test results to the way
the ground will respond at a specific limit state, including
effects of construction on its behaviour. Eurocode 7 also notes
the need to be aware that extreme values of strength may be the
critical ones if a failure mode could exploit them.
An alternative approach is also mentioned by Becker (1996)
and has been used in Swedish practice (Boverket, 1995). In this
approach, the “Characteristic value” is taken to be a simple
mean value, or best estimate of the actual value. The value of
applied partial factors is varied in a prescribed manner related to
the assessed reliability and variability of the data from which
the mean was derived, and its likely ductility. The author
understands that Swedish codes are changing towards a
definition which includes some of the uncertainty in the choice
of characteristic value, and is therefore closer to the Eurocode
definition.
With relatively small differences in approach, all these
documents share the aim of obtaining the most useful, rational
and transparent combination of objective test results with
engineering knowledge and experience, necessarily subjective
to some extent. A purely objective process would disregard
large amounts of human knowledge which are essential to
geotechnical design.
2.4 Can the various limit states really be treated separately?
In principle, limit state design requires separate consideration
and analysis of various different limit states, for example,
ultimate (or strength) and serviceability limit states. Separation
makes it easier to specify the requirements of design for each
limit state, and many authors, eg Becker (1996), have argued
that this is essential for the “rational” derivation of values for
partial factors, ie on a probabilistic basis. The split between
ultimate and serviceability has encouraged development of
testing methods and analytical tools for serviceability
calculations, generally aiming to provide ground stiffnesses for
calculation of displacements. The more detailed prescriptions
of the codes have generally been dominantly ULS, partly
because SLS criteria are the province of the client rather than
being requirements of society.
Parameter values and related calculation models giving
reliable predictions of displacement are often very difficult to
obtain. In contrast, those for mechanisms of plastic failure are
generally easier and more reliable; for example, the angle of
friction of a soil is much easier to obtain than its deformation
properties. Because of this, Eurocode 7 also allows SLS to be
covered by the strength-based (plastic mechanism) calculations
in some cases, as shown by the following extracts.
{2.4.1(4)P} If no reliable calculation model is available for a
specific limit state, analysis of another limit state shall be
carried out using factors to ensure that exceeding the specific
limit state considered is sufficiently improbable. Alternatively,
design by prescriptive measures, experimental models and load
tests, or the observational method, shall be performed.
{2.4.8(4)} It may be verified that a sufficiently low fraction
of the ground strength is mobilised to keep deformations within
the required serviceability limits, provided this simplified
approach is restricted to design situations where:
 B. Simpson et al. / State of the art Report: Analysis and Design 2877

• a value of the deformation is not required to check the
serviceability limit state;
• established comparable experience exists with similar
ground, structures and application method.
{6.6.2(16) [for spread foundations]} For conventional
structures founded on clays, the ratio of the bearing capacity of
the ground, at its initial undrained shear strength, to the applied
serviceability loading should be calculated (see 2.4.8(4)). If this
ratio is less than 3, calculations of settlements should always be
undertaken. If the ratio is less than 2, the calculations should
take account of non-linear stiffness effects in the ground.
{7.6.4.1(1)NOTE [for pile design]} For piles bearing in
medium-to-dense soils and for tension piles, the safety
requirements for the ultimate limit state design are normally
sufficient to prevent a serviceability limit state in the supported
structure.
The first of these extracts refers to limit states in general.
The second and third occur in sections on SLS, allowing
strength calculations which may be different from those
required for ULS to be used to demonstrate limited settlement.
It will normally be the case that this SLS condition is, by
inspection, more severe than the ULS condition which would
use the same type of calculation, so one calculation checks both
SLS and ULS. The fourth extract, referring to piles, specifically
states that the ULS calculations will normally cover SLS also;
again, one calculation covers both. In the European system,
actual values for partial factors are set by each nation, so each
must ensure that the values set comply with this note; the
application of this in developing the UK National Annex for
EC7 (BSI 2007) is discussed by Bond and Simpson (2009).
If the values of partial factors were derived by probabilistic
analysis, it would be confusing to mix SLS and ULS, as noted
by Becker (1996). However, in practice the values adopted in
codes to date are mainly chosen to reproduce existing
experience of successful behaviour. Since this implies success
both in ULS and SLS, it is clear that the factors being chosen
are adequate in most cases to provide for both, and it may be
very difficult to determine which of the two limit states actually
governs the factor values. Factors based on past successes
apparently cover all significant variables, to a degree which is
generally adequate. Besides uncertainly of loads and material
strengths, this includes calculation models, geometric
uncertainties and human errors.
In the opinion of the authors, statistical studies are useful to
code drafters in helping to allocate partial factors in appropriate
proportions between the parameters. However, when the
overall level of safety derived by modern codes is fixed by
calibration with past successful experience, actual magnitudes
may well cover more than one limit state and inevitably allow
for a greater range of uncertainties than the statistical variation
of the basic parameters. A lack of theoretical purity in approach
is probably of little concern to designers, provided codes are
clear about what limit states are covered by each of their
requirements, and in particular when it is necessary to carry out
specific calculations of deformation and displacement.
structural design in situations where unfavourable and
favourable loads tend to cancel in the expected state, leaving
only small working stresses, but where a slight increase in the
unfavourable loads could lead to proportionately very much
bigger increases of stress.
In November 1965, during severe wind conditions, three of
the eight cooling tower shells collapsed at the Ferrybridge C
Power Station in Yorkshire, UK (Figure 2.3). The Committee
of Inquiry into the collapse reported that wind conditions were
“very considerably lower than the probable maximum values”
and that the structural analysis had been carried out correctly.
They reported: “The important membrane stresses in the shells
of the towers are the resultant of compressive stresses due to
dead weight and tensile stresses due to wind uplift: the resultant
tensile stress, being the difference between two large quantities,
is sensitive to small variations in the wind loadings.” This
event was critical in the rejection of working state design and
development of load factoring, indicating the importance of
considering each independent load separately.
Ferrybridge was not a geotechnical failure, but all the loads
in any structure have eventually to be brought down to the
ground, so the same thinking applies to foundation and
geotechnical designs. Schuppener et al (2009) discuss the
examples shown in Figure 2.4. In each case, the forces
transmitted to the ground may differ appreciably from the most
likely working state due to relatively small changes in the
structure and its loading. In Figure 2.4a, small changes in
loading change the forces in the piles from tension to
compression. In Figure 2.4b, small changes determine whether
an anchor is needed or not at point A.
To avoid problems of this type, the drafters of modern codes
have taken the view, rightly in the opinion of the present
authors, that designers should be required to consider states in
which relatively extreme values of parameters take the structure
close to failure.
(a)

2.5 Review of the alternative approaches

2.5.1 Working state design

As noted above, in Working state design attention is directed to
the expected, desired state in which the structure is performing
successfully under its expected loading. Its safety is then
usually checked by requiring that the degree to which material
strengths or ground resistances are mobilised is limited. This
means that “permissible stresses” are not exceeded, so the
approach is also called Permissible stress design or Working
stress design (WSD).
Working state design has the advantage that the designer is
asked to consider states which are easier to imagine because (b)
they are likely to occur. Its weaknesses became apparent in Figure 2.3: Collapse of Ferrybridge cooling towers
2878 B. Simpson et al. / State of the art Report: Analysis and Design
(a)
(b)
Figure 2.4: Examples of balanced loads (after Schuppener et al 2009)
2.5.2 Introduction of partial factors
For ultimate or strength limit states, the requirements can
generally be resolved into a condition of the form Ed ≤ Rd. That
is, the design effect of loads must not exceed the design
resistance, the term “design” implying that all necessary factors
are already incorporated. This condition may be checked at
many different points in a structure, typically for example as a
thrust, tension, shear force or bending moment in a structural
member, or as a bearing pressure, disturbing moment or lateral
resistance in a geotechnical calculation. The term Ed is formed
by the combination of separate loads with their own factors and
the term Rd is calculated from characteristic values of ground
strength using theoretical or empirical relationships of varying
provenance and reliability. Significant differences of opinion
exist in deciding where in the process of deriving Ed and Rd the
factors should be applied.
Eurocode 7 effectively allows a very general equation:
γE E{γF Frep; Xk/γM; ad} = Ed (2.1)
≤ Rd = R{γF Frep; Xk/γM; ad}/γR
in which the main variables are Frep, Xk and ad – representative
loads (or “actions”), characteristic ground strengths and design
geometrical parameters (representative actions are derived from
characteristic actions using load combination factors, outside
the scope of this paper). The functions E{} and R{} represent
the derivation of the load effect from the individual loads and of
the corresponding resistance from the ground strengths and
other parameters. The factors are γE, γF, γM and γR on load
effects, individual loads, material strengths and resistances,
respectively, some of which may be set to unity. The material
strength term Xk is included in the load effect side of the
condition because ground strength may affect load effects such
as lateral earth pressures; similarly the load term Frep is included
on the strength side because in frictional soils loads may affect
strength – for example normal loads enhancing sliding
resistance. In North American usage, the reduction factor φR is
used in place of 1/γR.
Between the various groups working on code development,
opinions have diverged about which of the factors γ can most
usefully be set to unity, their effect being incorporated into
other factors. All developers aim to provide a system which is
robust and reasonably simple to apply. In the author’s
experience, the greatest complexities and lengthy debates arise
in design practice when the system proposed leads to obviously
unreasonable results; sometimes this occurs as a result of over-
enthusiasm for simplicity.
2.5.3 Factors applied to material strength or resistances
On the resistance side, the main debate is whether to apply
factors γM to material strength (cu, c′, tanφ′) or factors γR or φ to
derived resistances such as bearing capacity or lateral
resistance. Becker (1996) put the case for resistance factors as
follows:
“Although the factored strength approach may be considered
as being more elegant and sophisticated, the factored overall
resistance approach has a significant advantage over the
factored strength approach in that the derived resistance factor
reflects not only uncertainty in strength but also uncertainties
associated with the analytical models, site conditions,
construction tolerances, and failure mechanisms. The factored
strength approach alone does not capture all sources of
uncertainty in the calculation of resistance. There is a need to
also take into account the uncertainties stated above and others
such as the development of excess pore-water pressure and
stress-strain behaviour. The factored strength approach also
does not capture the true mechanism of failure when failure is
influenced by nonlinear soil behaviour. … The factored
resistance approach is similar to WSD and may be viewed as a
logical extension to WSD. Therefore, it would be familiar to
and, hence, better received by geotechnical engineers, which
would allow for a smoother transition from WSD to LSD for
foundation design.”
In Eurocode 7, both possibilities are allowed and the choice
is left to individual nations. There has been no general
agreement as to which approach is preferable or, indeed,
whether either approach is best in all circumstances. The main
argument voiced in favour of the resistance factor approach has
been the second main point made above by Becker, that this
more readily reflects existing practice based on single factors of
safety. However, some European countries diverged from that
system many years ago; examples include CIRIA Report 104
(Padfied and Mair 1984), the Danish code (Dansk
Ingeniorforening 1984), Norwegian codes (Det Norske Veritas
1993), and the British retaining walls code (BS8002 1994). It
would be difficult to imagine a single set of factors which could
adequately take account of “uncertainties associated with the
analytical models, site conditions, construction tolerances,
failure mechanisms, … excess pore-water pressure and stress-
strain behaviour”. The Eurocode approach is generally not
specific about which analytical models are to be used but
requires that they should be “accurate or erring on the side of
safety”; where necessary, an additional resistance factor (or
“model factor”) may be introduced to ensure this. The code
requires other uncertainties such as construction tolerances to be
considered specifically if they are significant and similarly
requires either that extreme pore pressures are considered or
that load factors are applied directly to them. In all systems,
minor variations in these secondary variables are implicitly
covered by the factors on the leading variables.
Simpson (2007) considered some of the advantages of a
material strength approach in geotechnical design, in particular:
• It facilitates consistent analysis of combined problems,
which are very common, involving, for example, a slope,
loaded by a structure, supported by a retaining wall, itself
supported by anchors and foundations. An example is
shown in Figure 2.5.
• Because the strength of soil is derived from friction, non-
linear, or disproportionate relationships between soil
parameters and resistances are common. In these
circumstances, it is important to check designs with factors
of safety applied to the basic strength parameters of the soil,
as explained fully in EN1990 {6.3.2}. Such non-linear
 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 2.5: Combined geotechnical and structural design situation
relationships occur, for example, in the derivation of
bearing capacity or of passive resistance from angle of
shearing resistance, as illustrated in Figure 2.6. (Kp is
2
plotted for δ/φ= ⁄3.)
• It can be used readily with both simple hand calculations
and more complex finite element calculations. Introduction
of resistance factors into finite element computations, which
essentially require overall equilibrium, has proved to be
difficult. In the authors’ view, demonstration of equilibrium
of complete systems is fundamental to good design practice.
Figure 2.6: Non-linear relationships between material property and
resistance.
Foye et al (2006 a,b) report studies intended to derive values
for resistance factors φR for design of spread foundations. Their
work is based on statistical studies of the variability of material
strengths, as measured by CPT and SPT, and does not
incorporate explicitly the other variables listed by Becker
(1996). Indeed, these would be very difficult to include since
statistical studies of the behaviour of spread foundations in
practice are not available. Starting from an assessment of the
coefficient of variation of angle of shearing resistance, ϕ′, they
aim to achieve constant reliability for bearing capacity and
conclude that ideally the factor φR on bearing capacity should
vary as a function of ϕ′. Figure 2.7 shows how their
recommended value of φR varies with “nominal” (equivalent to
“characteristic”) value of ϕ′ derived from CPT results. The
graph is plotted for a live/dead load ratio of 0.5, and four curves
are plotted for four values of reliability index β that they
consider to be relevant, with preference for β=3.0. The figure
2879
also shows the equivalent value of φR provided by a constant
factor of 1.25 applied to tanϕ′, a typical value of γϕ used in
EC7; adjustments have been made to allow for the differences
between load factors in the two approaches. (The plot for DA2
will be discussed later.) It can be seen that the constant γϕ
provides an equivalent φR having a similar variation with ϕ′ to
that proposed by Foye et al, and corresponds, according to their
analysis, to a reliability index β of between 2 and 2.5. The
precise value of β is dependent on the way the characteristic or
nominal ϕ′ is derived; Foye et al show different values when it
is derived from SPTs because greater uncertainty is involved.
Figure 2.7: Relation of φR to ϕ recommended by Foye et al (2006b)
The implications of this comparison for ULS can be seen in
Figure 2.8, which shows results obtained by Foye et al (2006b)
of the width of strip footings as a function of nominal angle of
shearing resistance ϕ′, obtained from interpretation of CPT
results. Also shown are calculations following EC7 DA1,
plotted against characteristic angle of shearing resistance. IN
both cases, the footing with B has been normalised in relation
to the soil weight density γ and the total design load ΣFd used by
Foye et al. They use a formula for the bearing capacity factor
Nγ taken from Salgado et al (2004) which is more conservative
than the EC7 formula, so calculations using the EC7 partial
factors are shown for both bearing capacity formulae in Figure
2.8. As might be anticipated from Figure 2.7, the two sets of
results are similar EC7 results are similar, with the EC7 results
somewhat less conservative.
An interesting feature of this work is that Foye et al based
their analysis on an assessed coefficient of variation of angle of
shearing resistance, which would most readily be represented by
a factor on tanϕ′ (or ϕ′ - it makes little difference), rather than
by a φR. The same implication was made in previous work by
Burland et al (1981) who developed a new method for
representing factor of safety in design of embedded retaining
walls. Again, they justified it by showing that it was equivalent
to a constant factor on soil strength, implying indirectly that
factoring soil strength is a sound way to proceed. It seems that
geotechnical engineers intuitively expect that a constant factor
on tanϕ′ is a good way to give a reasonably constant level of
safety. This may reflect the point noted above that when
strength is derived from friction, non-linear, or disproportionate
relationships between soil parameters and resistances are of
concern.
One problem of a material strength approach is that some
calculation methods derive bearing resistance, for example,
directly from results of in situ tests, without the use of a
2880 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 2.8: Footing dimensions according to Foye et al (2006b) and
EC7
parameter of material strength. If it is significant, this difficulty
could be overcome by having a resistance factor which varies as
a function of the test result, much as suggested by Foye et al.
The special case of pile design is considered below.
It was noted above that because factor values are inevitably
based on experience to a considerable extent it is impossible to
distinguish whether they are required by ULS or SLS
considerations. Bolton (1993) has argued that in many cases
SLS requirements can reasonably be accommodated by
checking the degree of mobilisation of material strength (also
Osman et al 2004, Osman and Bolton 2006). This is also
readily facilitated by factors on soil strength, relieving code
drafters, to some extent, of the need to make the difficult
distinction between requirements for ULS and SLS.
2.5.4 Factors applied to loads or load effects
The collapse of the Ferrybridge cooling towers, noted above,
illustrates the danger of combining loads too early in a
calculation. Equation 2.1 includes factors on both loads and
load effects. Eurocode 7 allows either to be used, but in either
case requires different factors on permanent and variable loads.
In general, it could be difficult to distinguish the effects of two
different loads after they have been combined into a single load
effect. The situations shown in Figure 2.4 are again noted as
examples.
This issue is particularly critical to the design of spread
foundations because calculated bearing capacity is very
sensitive to inclination and eccentricity of load. To augment
information culled from literature, the author has discussed this
with various experts. In the development of Japanese codes,
Shirato (2008) notes that factors are generally applied to loads
before combination, though the client body for railway
structures requires the use of unfactored loading, with all safety
factors on resistances. For the Canadian Highway Bridge
Design Code, Green (2008) notes that loads are to be factored
before being combined, so the inclination and eccentricity used
in the ULS calculation are based on the factored loads. Becker
(2008) notes that this question has been debated in North
America, without unanimous agreement. The AASHTO LRFD
Bridge Design Specifications (2008) uses factored loads to
compute eccentricity and Figure C11.5.5-2 shows that earth
pressure forces are factored before being combined for
calculation of eccentricity, stating “Abutments, piers and
retaining structures and their foundations and other supporting
elements shall be proportioned for all applicable load
combinations” The Specifications state “Bearing resistance
shall be investigated at the strength limit state using factored
loads” {11.6.3.2}. Despite this, for the calculation of load
inclination unfactored loads are used {eg 10.6.3.1.2a}; an
earlier edition gave the explanation “To preserve the angle of
inclination of the resultant, unfactored loads are used. When
factored loads are used, the failure surface beneath the footing
could be different than the one due to the applied loads, and the
result may become overly conservative.” Modjeski and Masters
Inc (2003) provide an example of this for a spread foundation
for a retaining wall designed to AASHTO Specifications.
The various National Annexes to EC7 take differing views
of this question, as described below.
2.6 Eurocode 7
EC7 requires that both ULS and SLS be considered. Most of its
text refers to ULS; text referring to SLS was discussed above.
For ULS, the main approach is based on use of partial factors,
but opinions in Europe differ about where and how these should
be applied. This is left to national choice, and the values to be
adopted for partial factors may also be varied nationally. Three
alternative “Design Approaches” have been developed,
combining partial factors in different ways; the factor values
proposed in the European document are shown in Figure 2.9.
Bond and Harris (2008) have summarised the adoption of the
three Design Approaches by the various countries in Europe, as
shown in Figure 2.10, for designs other than slope stability.
Most countries that have adopted DA2 for other purposes have
decided to use DA3 for slope stability.
In Design Approach 1 (DA1), two “combinations” of partial
factors are specified, and the design must be shown to
accommodate both combinations (Figure 2.9). Essentially, they
are used in the same way as load combinations, but the concept
is extended to include material strengths and resistances. Partial
factors are generally applied to either loads (before
combination) or ground strengths (before calculation of
resistances), though with some exceptions. In countries that use
DA1, the factors on ground materials and strengths are
generally set to 1.0 in Combination 1; the factors adopted in the
United Kingdom are shown in Figure 2.11. For design of piles
and anchors, factors are applied to resistances rather than to
material strengths, for two main reasons: (a) the designs often
involve load testing, which leads directly to resistances, and (b)
because the construction process may change the ground
strength, a major uncertainty exists in deriving resistances from
the strength of the ground before construction; the resistance
factors can be seen as factors on this process (or “model”) or on
the strength of the ground at the interface with the pile or
anchor. There are some situations in which factoring loads at
source leads to unreasonable situations, especially in the design
of retaining structures. For these, EC7 allows the factors to be
applied to the effects of the loads, and this is used where
appropriate in “Combination 1” of DA1.
In DA2 (Figure 2.9, partial factors are applied to loads and
to ground resistances. In a variant of DA2, DA2*, the
equilibrium calculation is carried out using unfactored
(“representative”) loads, and the factors are applied to derived
load effects. It has been found that DA2 and DA2* are
unsuitable for slope stability problems, so most countries which
have adopted DA2 use DA3 for slope stability.
In DA3, factors are applied to material strength and to loads
simultaneously, in contrast to the two-combination approach of
DA1 in which they are applied to the two separately and the
results compared. DA3 has been adopted mainly for slope
stability problems only, though a few countries propose to use it
for foundations and retaining walls, with factors quite different
from those shown in Figure 2.9 in most cases.
Simpson (2007) has emphasised that in many designs
involving slopes, retaining structures and foundations ground
and structures have to perform together and if a failure occurred
 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 2.9: Factors proposed by CEN for the three Design Approaches
it would involve both of them The main advantages of DA1
were stated to be:
a. Consistent design of complete ground-structure systems
with the same two sets of factors.
b. More consistent reliability across a wide range of
geotechnical problems including foundations, slopes and
retaining structures.
c. Factoring materials mainly at source, better control of safety
is obtained in frictional materials which give non-linear
relationships to resistances.
d. By factoring loads mainly at source, better control of safety
in cases where loads tend to cancel each other.
e. Ready use with finite element methods.
The main reason for adoption of DA2 has usually been stated
to be that it is more like previous practice and produces results
consistent with previous practice. This has been particularly
used in support of DA2*.
Figure 2.10: Adoption of the three Design Approaches by the various
countries in Europe (after Bond & Harris 2008).
2881
The author has found that the results of DA1 and DA2 are in
most cases fairly close, though DA2* may produce more
economic, less safe designs in some cases. As noted above,
previous practice varies between countries. For the design
problem shown in Figure 2.12, Orr(2005) found similar results
for DA1 and DA2. DA1 Combination 2 gives the design length
for DA1, which is 4.73m, and the design maximum bending
moment is 163 kNm. The design length for DA2 is 4.69m,
which is marginally less than the DA1 length, while the
maximum design bending moment is 177 kNm, which is
slightly greater than DA1. There is no application of DA2* in
this example, and DA3 is equivalent of DA1 Combination 2.
It is hoped that some convergence of these approaches will be
achieved in the future. The combined used of DA2 and DA3,
adopted by some countries, is similar in many ways to the two
combinations of DA1, suggesting some progress towards a
common approach.
2.7 Some specific examples
2.7.1 Spread footings
For spread foundations with vertical loading, EC7 Design
Approaches 1 and 2 generally give similar results. For example,
the factors in Figure 2.9, Orr (2005) calculated widths of
foundations within 10% of one another for a typical example. It
is found, however, that the sizes of footings computed generally
imply high bearing pressures, even for the unfactored case, and
it is likely that serviceability checks will dominate the design,
leading to bigger footings. Footings designed to ASHHTO
(2008) strength limit state would be considerably larger than
required by the EC7 ULS. As seen in Figure 2.7, EC7 Design
Approach 1 follows the trend proposed in recent studies by
Foye et al (2006b), while Design Approach 2 allows for less
uncertainty of behaviour at high angles of shear strength.
For combinations of vertical and horizontal loads, however,
DA1, DA2 and DA2* may differ appreciably because the load
factors are applied at a different point in the process. Figure
2.13 shows an example published by Vogt and Schuppener
(2006) which provides an illustration of the effect of factoring
loads before or after they are combined. This involves the
calculation of the width of a strip footing subject to inclined
eccentric loading, based on ULS bearing capacity calculations.
A range of the characteristic value of the horizontal force
HQ,k is considered. Figure 2.14 shows that Combination 1 of
DA1 requires two separate action combinations to be
considered, applying 1.5 to the variable horizontal action, and
2882 B. Simpson et al. / State of the art Report: Analysis and Design
Figure: 2.11: Factors adopted by the United Kingdom for Design Approach 1.
Figure 2.12: Embedded retaining wall considered in the Dublin
workshop of 2005
Figure 2.13: Inclined, eccentric loading on a footing
1.35 to the permanent vertical action if it is unfavourable (red
dashed arrow), or 1.0 if it is favourable (red solid arrow). This
is consistent with EN1990 and the Eurocodes for structural
design, and in principle it is also required for DA2 and DA3.
Figure 2.15 shows similar diagrams for Combination 2 of
DA1 and for DA2 and DA2*. In DA2*, for consideration of
bearing capacity, the horizontal and vertical loads are combined
into a single resultant before the load factors are applied, so the
inclination and eccentricity of the design action effect takes no
account of the separate factors on permanent and variable
load, or of the possibility that the permanent vertical load is
favourable.
The results of calculations of the required footing width for
varying ratio Hk/Vk are shown in Figure 2.16, using the bearing
capacity equations in Annex D of EC7. Results for DA1 and
DA2 are fairly similar, and DA3 is more conservative, as
expected. However, DA2* gives a markedly less conservative,
and so less safe design. In Figure 2.17, the factors of safety
available on the single quantities (a) γϕ on tanϕ′ and (b) γQ on
HQ have been calculated using the footing widths obtained by
DA2*. This shows that DA2* is equivalent to a single factor of
safety slightly greater than 1.2 on tanϕ′ or a variable factor on
HQ which falls to about 1.17 at HQ,k/VG,k of 0.4, the
approximate limit of this ratio for sliding (EC7 allows δ=ϕcv′ for
concrete case on the ground). As the only safety factor in the
calculation, a value of 1.17 on the variable load is remarkably
low; a factor of 1.5 is generally applied to variable loads..
The proponents of DA2* argue that the results it obtains are
in line with previous experience. However, the author submits
that: (a) it is unlikely that there is a significant database of
footings which have, in practice, been loaded to the high H/V
ratios considered here (though it is very important that footings
which have to be designed for high ratios are reliable); (b) the
approach taken for bearing capacity is inconsistent with that
taken for sliding, which is likely to cause confusion; (c) the
approach taken for bearing capacity is inconsistent with that
taken for structural design, which is also confusing; (d) as noted
above, as the only safety factor in the calculation, a value of
1.17 on the variable load is remarkably low.
In practical use, design rules are often pushed to extremes
not anticipated by code drafters. For this problem, it is
desirable to check what happens if the lever arm of the
horizontal load is larger. It is assumed here that structural
designers would factor the vertical and horizontal actions
independently. Figure 2.18, for a force at 10m above the base,
shows that at a relatively modest ratio Hk/Vk= 0.13 the
eccentricity of the resultant load used in structural design
exceeds the half width of the footing derived by DA2*,
implying that the resultant load passes outside the base of the
footing, which cannot give equilibrium. This will clearly cause
consternation and confusion to the structural designer, and
illustrates the inconsistency of DA2*.
 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 2.14: ULS design resultant actions derived for DA1.
Figure 2.15: ULS design resultant actions derived for DA1, DA2 and
DA2*
Figure 2.16: Footing widths calculated for ULS design
In contrast, Figure 2.19 shows that for DA1 the resultant
load always lies within the width of the footing; Combination 2
is more critical than Combination 1, and so determines the
width. Figure 2.20 shows that DA1 even accommodates a
much more extreme lever arm for the horizontal load of 100m.
This is guaranteed since the geotechnical design is checked for
the same loading as the structure, consistently, in Combination
1, and for this extreme case it is Combination 1 which
determines the footing width. For such an extreme case, much
of the width of the footing might be redundant and the structure
could be replaced by an A frame, but all the considerations
noted above would still apply.
2.17: Total factors of safety implied by DA2 on tan ′ and HQ, taken
singly.
2883
Figure 2.18: Eccentricities in DA2 and structural design, for 10m lever
arm
Figure 2.19: Eccentricities in DA1 and structural design, for 10m lever
arm
Figure 2.20: Eccentricities in DA1 and structural design, for 100m lever
arm
2.7.2 Piled foundations
Traditional pile design has generally used overall factors of
safety in the range 2 to 3, or even up to 4 in the case of tension
loading. Often the lower end of this range is used when the
design is verified by load testing; and often larger values are
applied to base resistance than to shaft resistance. A major
reason for the use of relatively high factors has been that they
ensure serviceability as well as safety against ultimate failure.
It is therefore necessary for developers of modern codes to
decide whether they will perpetuate this situation or try to
separate ULS from SLS. If a separation is attempted, it will not
be possible, in many cases, to calibrate factors needed for ULS
against existing practice, and SLS will often dominate the
design, at least for larger diameter piles.
EC7 requires a “characteristic” pile resistance which is a
“cautious estimate”. In practice this can be derived from
calculation based on ground testing or from load testing of piles,
or a combination of the two. Indeed, EC7 specifically requires
that calculation methods must be verified by testing, and test
results must be checked by calculation {7.4.1(1)}, but it does
not clarify how information from these two sources can be
combined when both are used together. It also states that “For
piles bearing in medium-to-dense soils and for tension piles, the
2884 B. Simpson et al. / State of the art Report: Analysis and Design

safety requirements for the ultimate limit state design are or ground strength, such as the use of piles or anchors used to
normally sufficient to prevent a serviceability limit state in the hold down the slab as shown in Figure 2.22(b).
supported structure” {7.6.4.1(2)}. Thus for piles, in contrast to
other design elements, the code implicitly requires that factors
for ULS design be chosen so that SLS will be satisfied. Hence
the overall factors of safety adopted should probably be similar
to previous practice.
Several European countries have reported that the partial
factors proposed by EC7 imply overall factors of safety much
lower than they would traditionally have used. The code is
specific about the process of deriving design resistances from
load tests, applying “correlation factors” ξ to the mean and
lowest test results. Much less detail is given about calculation
of resistances from ground strength tests, however. This means
that each country has some freedom to specify in rather more
detail how pile resistances will be calculated, and the values of
partial factors may also be varied nationally, as for other design (a)
elements.
Bond and Simpson (2009) describe the background to the
derivation of the factor values required by the UK National
Annex to EC7 (BSI 2007). The partial factors γ have been
increased compared with the values in the base version of
EN1997-1 (see Figure 2.9); the correlation factors ξ have also
been increased and additional “model factors” have been
imposed in the derivation of characteristic shaft and base
resistances from characteristic ground strengths. Both the
model factors and partial factors γ have been varied according
to the extent of load testing which supports the calculations.
The overall effect for ULS calculation is shown to be consistent
with previous design, and the characteristic values of resistance
(b)
are suitable for serviceability calculations as, for example, in
settlement reducing piles which might be loaded to their
ultimate capacity in the working state. Example calculations are
provided by Bond and Simpson (2009).

2.7.3 Water pressures

Application of factors to water pressure is fraught with
difficulty and can easily lead to unreasonable situations. Orr
(2005) reported calculations for the situation of potential
hydraulic heave shown in Figure 2.21. He found that the
calculated allowable height of water H could vary from 2.78m
to 6.84m due to application of the same factors, taken from
EC7, at different points in the calculation. (c)
Figure 2.22: (a, b) ‘UPL’ and (c) ‘HYD’ situations in EC7.

Modern codes tend to provide a single factor for “dead

Figure 2.21: Hydraulic problem considered in the Dublin workshop of
2005
In design of structures in the ground, water pressure may
constitute an additional load to be considered along with others,
to be accommodated by strength requirements in structure
and/or ground. In addition to this, two particular situations can
be identified in which water pressures are principally balanced
by other loads (weight of structures or ground): uplift failure
and hydraulic failure, termed UPL and HYD in EC7, as
illustrated in Figure 2.22. EC7 provides factors of safety to be
used in checking these limit states, but it is not clear about
where in the calculation they should be applied, leading to the
confusion noted by Orr. In practice, these states are often
complicated by the involvement of some element of structural
loads” or “permanent actions”. It is arguable that static water
pressure in the ground is a permanent action. In some cases,
design can proceed, apparently reasonably, with a factor (eg
1.35) applied to water pressure, but in other cases applying a
factor to water pressure is unreasonable. For example, Figure
2.23 shows a deep basement or shaft in ground with a high
water table; increasing water pressure in the ground near the
base of the shaft by a significant factor (eg 1.35) would
represent a physically impossible situation and would constitute
an unreasonably onerous design case.
The AASHTO code {10.6.3.1.1} requires that “bearing
resistance shall be determined based on the highest anticipated
position of groundwater level at the footing location”, but it
does not apply factors to water pressures for foundation or
retaining wall design, despite factoring effective earth pressures.
This appears to imply that in a situation where ground water
pressure is dominant the design would rely almost entirely on
the resistance factors in both the ground and the structure.
Eurocode 7 requires the designer to consider two situations.
For ULS, the design should use “the most unfavourable values
that could occur during the design lifetime of the structure”,
whereas for SLS “the most unfavourable values which could
occur in normal circumstances”. It also allows that design
 B. Simpson et al. / State of the art Report: Analysis and Design
water pressures for ULS could be derived by applying a partial
factor to characteristic water pressure or by raising the water
level. Figure 2.24 compares water pressures for a 10m high
wall with a 2m difference between these two levels. Starting
from the “characteristic” water pressures (line A), it can be seen
that merely applying a factor (obtaining line B) may
underestimate the effect of raising a water level (line C); this is
particularly the case when thrusts or bending moments are
considered. On the other hand, factoring water pressures which
are already “the most unfavourable values that could occur
during the design lifetime of the structure” (obtaining lines D or
E) seems unreasonable from both probabilistic and physical
points of view.
Figure 2.23: Deep basement or shaft in ground with a high water table
Figure 2.24: Water pressures for a 10m high wall
The UK National Annex to EC7 (BSI 2007) points designers
away from simple factoring of water pressures, though it leaves
the option open. It says that standard load factors “might not be
appropriate for self-weight of water, ground-water pressure and
other actions dependent on the level of water”, and “the design
value of such actions may be directly assessed in accordance
with” the EC7 principles noted above, or alternatively “a safety
margin may be applied to the characteristic water level”.
In contrast, Bond and Harris (2008) provide two arguments
in favour of factoring water pressures when designing retaining
structures: (a) structural engineers have traditionally applied
partial factors to retained liquid loads and ground water
pressures, and (b) it is common for geotechnical engineers to
perform calculations using unfactored parameters, including
water pressures, and then to apply a factor of safety to the
resultant structural effects, giving the same result as factoring
both the effective earth pressures and water pressures.
Provisions for safety in relation to water pressure, both free
water and ground water, remain under debate within the EC7
Maintenance Group. Given such uncertainty, making more than
one check, effectively a parametric study, may be advisable.
For EC7 Design Approach 1, the author has found that for most
situations the following strategy gives reasonable results,
though critical review is always necessary:
2885
1. Apply DA1-1 using reasonably cautious water levels (most
unfavourable in normal circumstances) and apply 1.35 to
structural load effects - bending moments etc.
Alternatively, in some cases, it is necessary to apply this
factor to all permanent actions, including water pressures.
Particular caution is needed if non-linear behaviour of
structures is expected.
2. Apply DA1-2 using worst credible water levels (most
unfavourable in design lifetime), unfactored.
2.7.4 EQU
Situations in which independent loads tend to cancel each
other’s effects were discussed above when considering working
state design. In a partial factor approach, the design is generally
made robust by applying different factors to the two loads. A
further problem is sometimes encountered when two loads
which are not independent, but arguably two parts of the same
load, have a cancelling effect.
Eurocode “Basis of design”, EN1990, considers that such
loads come from a “single source”, and notes: “For example,
all actions originating from the self weight of the structure may
be considered as coming from one source; this also applies if
different materials are involved.” When there is good reason to
believe that the two loads must occur together, it seems
unreasonably severe to factor them separately. Nevertheless,
some allowance for variability of loads also appears necessary.
EN1990 terms this situation the “EQU limit state” and requires
a check by applying small load factors, generally 1.1 and 0.9, to
the two parts of the single source load.
Schuppener et al (2009) have debated the significance of
EQU to geotechnical design, presenting several alternative
approaches without an agreed conclusion. In Figure 2.4(a), the
approach to this issue determines the design bending moment in
the column and changes the design loading in the piles, possibly
putting one of them into tension; in Figure 2.4(b) it affects the
design load in the anchor. The author’s opinion is that “the
whole system, both structure and geotechnics, should be
designed for all the relevant design situations or load cases”;
alternative approaches are inevitably more confusing and run
the risk of leaving gaps between design cases, in which either a
critical situation is not accommodated or design appears
impossible.
2.8 Economy and sustainability
Codes influence both the safety of constructions, probably their
primary aim, and their economy of constructions. With modern
appreciation of the need to reduce usage of materials and
energy, it is clearly desirable to produce more economic
designs, which might be assisted by reductions in values of
partial factors, where possible. It is commonly stated that
geotechnical failures, ultimate or serviceability, are rarely
caused by having factors of safety slightly too low. It is also
reported in Eastern European countries that the factors proposed
in EC7 (as published by CEN) lead to designs which are
uneconomic compared to their previous practice, which they
found satisfactory. This suggests that the proposed factors
might be reduced, and so a proposal has been put to CEN to
investigate further the designs used in Eastern Europe, with the
aim of reducing the proposed factors. If it is possible, it will be
necessary to compare records for safety and durability between
societies adopting different designs. Hopefully, common
international codes will provide a framework in which this
might be achieved.
Individual factors must be reduced with care since factors on
one variable contribute to the overall robustness of the design,
perhaps giving unanticipated benefits . For example, the public
enquiry into the Nicoll Highway collapse in Singapore (Magnus
et al, 2005) noted that one of the contributory causes was factors
of safety lower than code requirements. While this was not the
prime cause of the failure, Simpson et al (2008) have argued
2886 B. Simpson et al. / State of the art Report: Analysis and Design
that use of standard factors would probably have prevented
collapse, mitigating the effects of more serious deficiencies.
Further more, while reduced factors might not lead to collapse,
they could result in more durability problems in later years. It is
therefore very important that large scale surveys of performance
are undertaken as an aid to introducing reductions in code
factors.
Both economy and safety should be aided by more realistic
evaluation of the necessary resistance of structures and ground.
To this end, the Japanese development of Performance Based
Design aims to provide a framework which incorporates limit
state design and is particularly applicable to seismic situations.
Probably the biggest influence geotechnical codes can have
on both safety and economy, however, is to emphasise the
importance of good geotechnical process, desk study, ground
investigation and choice of construction type. Calculations and
factors, necessary for analysis and design, are important, but
secondary. To quote EC7: “It should be considered that
knowledge of the ground conditions depends on the extent and
quality of the geotechnical investigations. Such knowledge and
the control of workmanship are usually more significant to
fulfilling the fundamental requirements than is precision in the
calculation models and partial factors.”
3 DEEP FOUNDATIONS
3.1 Introduction
The 2005 ICSMGE Osaka conference included a State-of-the-
Art paper by Mandolini et al (2005) titled “Pile foundations:
experimental investigation, analysis and design” and a further
paper by Randolph et al (2005) titled “Challenges of Offshore
Geotechnical Engineering”. In Mandolini et al (2005), the
authors reviewed both single pile behaviour and pile group
behaviour, considering both experimental results and analysis.
The paper by Mandolini and his co-authors remains a thorough
review of these aspects of pile design and analysis. In Randolph
et al (2005) a summary of offshore ground investigation is
presented along with design and construction aspects of piled
foundations and spread footings (suction caissons, gravity
structures etc). Four years further on, the ISSMGE committee
for the Alexandria conference has asked for this short
contribution on deep foundations as part of a more general
paper on design and analysis; the committee suggested that
large diameter or “mega” piles could be discussed. This
contribution provides an update on design processes for single
piles in soil and rocks using design processes from both
offshore and onshore developments with bias towards large
diameter piles and vertical loading.
In addition to the sections on load-carrying capacity,
comments on sustainable aspects of pile design are provided in
view of the recent progress in this area and the lack of inclusion
in previous ICSMFE state-of-the-art reports.
Prior to presenting a limited number of pile design methods
it is worthwhile setting the scene for acceptable design.
Eurocode 7: Geotechnical design - Part 1: General rules (BSI,
2004) provides reasonable requirements for acceptable design:
The design [of piles] shall be based on one of the following
approaches:
• [Method 1] the results of static load tests, which have been
demonstrated, by means of calculations or otherwise, to be
consistent with other relevant experience;
• [Method 2] empirical or analytical calculation methods
whose validity has been demonstrated by static load tests in
comparable situations.
Within this report attention is paid to the second of these two
methods where pile load test data is used to develop design
methods. While the first method (design by site specific pile
test data) is not presented herein, it clearly remains a valid
approach provided measured pile resistances can be justified by
calculation methods derived in keeping with Method 2.
3.2 “Mega” piles: an introduction to vertical design
Where appropriate, large diameter piles are used in onshore
construction for infrastructure projects or high-rise building
development both of which generate high loads. The upper
limit of such piles is typically 1.5m in diameter, occasionally
2.1m or 2.4m in diameter, due to limitations on the size of
construction plant. In offshore locations “mega” piles with
diameters in excess of 5m are now possible owing to larger
construction plant and the economics of construction (speed of
construction is critical to economy off-shore). The design of
such large diameter piles is presented below with a bias towards
developments in offshore design practice.
Traditional methods of pile design rely heavily on
observation leading to empirical equations for assessment of
axial (geotechnical) capacity. This is still true, even though, as
pointed out by Randolph (2003a), the scientific understanding
of soil behaviour in general and around piles is continually
improving. Such improvements in understanding allow
observed pile behaviour to be better explained, thereby allowing
for better design. Design based on empiricism is nevertheless
the norm for assessment of pile capacity under axial loading,
although theoretically based equations are sometime used for
assessment of base resistance in some instances.
This general summary of the current state of design is
illustrated by the much used API (2000) “Recommended
Practice for Planning, Designing and Constructing Fixed
Offshore Platforms” document where much of design guidance
relies on empirically assessed resistance similar to Equations
3.2 and 3.3 below. (It is understood that a revised version of the
API document will be released in due course which will be
updated to included much of the work referenced below for
driven piles offshore.) The following review of pile design
shows that while empiricism remains at the forefront of pile
design, soil mechanics theory is continually developing and
providing important insights into the assessment of pile capacity
and behaviour. Soil mechanics should be seen as a route to
improving our understanding of pile behaviour and, with
recourse to best practice design methods, to informing ground
investigation techniques.
In the following, care is taken to present equations and data
for ultimate conditions. Where “allowable” or working
capacities are quoted, these are clearly identified. When
choosing the base soil parameters (e.g. characteristic values in
the Eurocode design) the designer must accommodate the basis
of the code adopted. For example, an average value of shear
strength may be appropriate to one code while in another the
parameter may need to be a cautious estimate. Failure to
understand the basis of the code will result in inappropriate and
potentially unsafe designs.
In assessment of the ultimate limit state (ULS) many
conventional design processes protected the serviceable limit
state (SLS) for conventional structures by limiting allowable
working load to a percentage of the skin friction (e.g. skin
friction / 1.2, in addition to a factor of safety on total capacity)
or by ensuring mobilisation of base resistance is small. As
codes become more sophisticated and as construction extends
beyond historic limitations (e.g. large diameter piles), there is a
greater need to assess pile settlement as part of the design
process with attention paid independently to the rate of
mobilisation of shaft and base resistances. Clearly where there
is direct comparable experience (loading, foundation size and
ground conditions) the requirement for explicit assessment of
settlement is reduced. Where such comparable experience is
absent then explicit assessment of settlement should be the
norm.
 B. Simpson et al. / State of the art Report: Analysis and Design
As a final introductory comment, no discussion on the
structural capacity or durability of piles is provided herein.
Such considerations are clearly required for a design in practice
and should accommodate applied actions and construction
tolerances as well as changes to the structure or dimension of
the pile with time (e.g. corrosion of steel piles).
3.2.1 Pile design based on offshore experience: soil
The following paragraphs present a summary of offshore pile
design for silica sands and clays. Comment is made on pile
capacity in carbonate soils in a following section.
Conventional design of piles usually adopts an equation as
follows:
Q = (Qb + Qs)= (Abqb + •A
Asqs) (3.1)
Where Q is pile resistance and q is limiting or failure stress,
A is area and subscripts b and s are base and shaft respectively.
The method for calculation of qb and qs has, in the past, been
relatively simple and commonly took the form of:
In sand:
Q = Abqb + •A
2
Asqs = ¼ D Nq 'v + D •(K 'v tan )dL (3.2)
and in clay:
Q = Abqb + Asqs =
2
¼ D Nc su base + D •(( su L)dL (3.3)
where the terms are as follows:
K earth pressure coefficient on the pile shaft
angle of interface friction (often expressed as a fraction of
the shear strength, e.g. 2⁄3 ϕ)
shaft adhesion factor applied to su
D pile diameter
L length for assessment of pile shaft friction
su undrained shear strength of the soil (clay). Subscripts
“base” and “L” illustrate the likely different values in su that
will be used in the calculation at the base and incrementally
along the shaft.
These equations are populated with a combination of
theoretical (Nq, Nc) and empirical (K, Į) parameters. When
using such equations there is significant guidance on the choice
of the design constants and limiting stresses for situations
within bounds of comparable experience. Such guidance has
been distilled into books such as Poulos and Davis (1980) and
Fleming et al (2008) among others. Whilst these equations and
their empirical parameters have come about from experience
and remain of practical importance, they do not provide
adequate confidence for situations where there is no comparable
experience. Such situations include new construction
techniques, sites with no previous pile performance data,
changes in pile dimensions compared to test data and where pile
construction is in new strata. Taking these conditions in turn:
• Where new construction techniques are adopted there is
little option but to carry out high quality instrumented load
tests to investigate pile behaviour. The results of such tests
should be interpreted so as to distinguish between pile shaft
and base behaviour (load settlement response). The results
should also be assessed against the best available
calculations in order to have confidence of the validity of
test data. Added confidence is gained as more piles are
tested, a process formalised by Eurocode 7 Part 1 (BSI,
2004).
• Where a new location is being developed, an understanding
of the soil is critical with, as a minimum, assessment of
geomorphological and geological situations (ground model),
mineralogy and soil consistency (relative density, strength
and stiffness data) being required to correlate to other
locations.
2887
• Where pile size is larger than test pile diameters, then
interpretation of test data must allow for the effect of
diameter on base and shaft capacities and overall settlement
to derive ULS and SLS conditions. As noted in sections
3.2.2 and 3.2.5 below the measured pile shaft resistance
(stress) varies as pile diameter increases thereby not
allowing linear extrapolation from small to large diameter
piles when assessing shaft capacity.
• Finally, where piles are constructed in new strata, the
behaviour of the strata during pile construction and
subsequent loading must be understood.
As an example of the above comment on soil strata, pile base
capacity calculated using a constant value of ϕ' and a
recognised bearing capacity factor (Nq ) such as Berezantzev et
al (1961) will fail to recognise the subtleties of soil behaviour at
varying stress levels as illustrated by Bolton (1987), Fleming
(1992) and as extended by Cheng (2004). In the context of pile
base bearing capacity, higher levels of confining stress result in
a reduction of the ability of the soil to dilate. This prevents the
peak angles of friction seen on samples tested at low effective
stress being mobilised. This observation is valid for siliceous
sands but becomes critical when softer soil grains are involved
due to early onset of particle crushing and further suppression of
dilatancy (Klotz and Coop 2001).
3.2.2 Offshore: Driven piles – a CPT approach
The environment in which offshore piles are constructed
combined with the requirement to carry large loads has resulted
in offshore piles being significantly larger than their onshore
counterparts. Driven tubular piles installed offshore can be 5m
and larger in diameter, dwarfing typical construction onshore.
The viability of such large piles is linked to progress in pile
installation technology and to the research into axial pile
behaviour resulting in new design methodologies as referenced
below. Whilst these new design methods are driven by the
requirements of offshore construction, they can equally well be
used to inform onshore pile design when similar construction
techniques are used and ground investigations carried out. They
may also be used as starting points for development of methods
appropriate to onshore practice. The design methods typically
use (sea-bed driven) CPTs to provide a profile of soil resistance
together with laboratory testing to provide further parameters
(density, interface strengths, magnitude and variation of Ko
etc.). Where CPT data is not continuous appropriate averaging
techniques must be used to generate a continuous profile.
The UWA-05 (University of Western Australia 05, Lehane
et al., 2005) method is used to illustrate the developments in
pile design in siliceous sand, albeit, as noted by Monzon (2006),
alternatives exist in the form of ICP-05 (Jardine et al) and NGI-
05 (Clausen et al, 2005). The UWA report deals with piles
installed (driven) in silica sands only. All methods mentioned
above differentiate between closed and open ended driven piles
owing to the different strain fields which develop during
driving.
Base resistance for close ended driven piles in sand is related
to CPT cone resistance qc as follows:
Qb = Ab qb0.1c (3.4)
qb0.1c = 0.6qc ave (for jacked piles the factor 0.6 can be increased
to 0.9)
qc ave the average value of qc over a distance 1.5 times the pile
diameter above and below the pile tip (further reference
is provided in Xu and Lehane, 2005).
The subscript 0.1c denotes the mobilised resistance at a
settlement of 10% (0.1) of the base diameter for a closed ended
(c) driven pile; the equation is clearly empirical.
2888 B. Simpson et al. / State of the art Report: Analysis and Design

For open ended piles experience (Lehane and Randolph,

2002) shows that in siliceous sand, piles loaded under static
conditions tend to plug when the height of plug inside the pile at
the end of driving is greater than 5 times the pile internal
diameter (5Di). This observation was made on piles of up to
1.5m diameter, care must be exercised for piles with greater
diameters and additional research is required to allow general
adoption of this observation. The result of this is that shaft
friction is generally linked to the external face of the pile only
and base capacity is modified as follows:

Qb = Ab qb0.1o (3.5)

where
Figure 3.1 Relationship between Effective Diameter, pile internal
qb 0.1o ª
2
§ D *· º (3.6) diameter and pile external diameter.
= «0.15 + 0.45¨ ¸ »
qcave ¬« © D ¹ ¼»

The subscript “0.1o” denotes the mobilised resistance at a

settlement of 10% (0.1) of the base diameter for an open ended
(o) driven pile. While it is assumed that the pile plugs, the open
ended pile “capacity” is less than that of a closed ended pile due
to the higher stiffness exhibited by the closed ended piles
compared to an open ended pile. This observation in stiffness is
a common thread seen in other empirical based design methods
as in Section 3.2.3 below. It is also consistent with previous
comments in Section 2.6.2 where partial factors for ULS pile
design generally provide a degree of comfort for SLS
considerations. D* is the effective diameter of the pile’s base
and is defined as (all pile diameter dimensions in metres):
D * §¨ ­° ª D º 0.2 ½°§ D 2 · ·
0.5
(3.7)
= 1 − min ®1, « i » ¾¨¨ i2 ¸¸ ¸
D ¨© °̄ ¬1.5 ¼ °¿© D ¹ ¸¹
For open ended piles qc ave is the average value of qc over a
dimension of 1.5 times D* above and below the pile tip and not
the external diameter, D, as for close ended piles. In Equation
3.7 the “min” value is referred to as the “final filling ratio”
FFR) and is, in the absence of direct measurement of the plug
length inside the pile, an approximation allowing for the plug
length. It should be noted that this FFR approximation only
holds good for situations where the pile tip has penetrated to at
least 5 times the internal diameter of the pile into the sand layer.
Where this is not the case, assessment of non-plugged pile
capacity will be necessary.
Value of D*/D are presented in Figure 3.1 for Di /D in the
range of 0.98 down to 0.0. Di /D = 0 for a closed ended pile and
0.98 for a thin wall pile. The limiting conditions occur when
the internal diameter equals 1.5m as can be seen in Equation 3.7
above. Figure 3.2 shows the variation in the ratio of end
bearing stress (average over the full cross sectional area) and
cone resistance in the vicinity of the pile tip with pile diameter
for a range of Di/D values; for closed ended piles the ratio is
0.6, falling to less than 0.18 for relatively thin walled tubes.
As would be expected, for a zero dimension of the internal
diameter Di (i.e. a closed ended pile), the value of Qb is the same
as a closed ended value in Equation 3.4 above.
Assessment of pile shaft capacity for open and closed ended
piles has been combined into a single method in siliceous sands
(Lehane et al, 2005b). This is possible due to the tendency of
open ended piles to plug during static loading thus negating the
need to assess skin friction on the inside of the piles tube..
While the method of assessment of pile shaft capacity is the
same for open end closed ended piles, closed ended piles
mobilise high resistances due to higher radial stresses acting on
the pile shaft resulting from the displacement nature of these
Figure 3.2 Relationship between end bearing stress on full pile
diameter and average cone resistance at pile tip with pile diameter as a
function of Di/D.
piles. Measured pile shaft friction is linked with cone resistance
and the fatigue induced at the interface between pile and sand
by pile driving as well as the interface friction (sand to pile) and
radial effective stress acting on the piles shaft due to the
displacement nature of the pile. The empirical correlation
between pile shaft friction qs and cone resistance qc is expressed
in terms apparently dissimilar to the shaft component in
Equation 3.2 above. The term Kı'v ( = ı'h ) in Equation 3.2 is
replaced by a function of qc which accounts for horizontal
effective stress working on the pile shaft. The horizontal
effective stress is a result of the post-installation conditions at
the pile face combined with the dilation of the soil adjacent to
the pile which is mobilised during pile loading. The latter term
can be compared to the enhanced resistance derived in a
constant normal stiffness shear box test, where dilation results
in an increase in confining stress and therefore shear resistance.
For large diameter piles the component of shaft friction derived
from dilatancy of the soil at the pile interface is relative small
allowing a simpler expression for shaft friction to be presented
(Lehane et al, 2005b):
ª § h ·º
−0.5
(3.8)
× tan δ cv
0.3
qs = a.qc Ars × «max¨ 2, ¸»
¬ © D ¹¼
qs = 'h x Fatigue factor x Coefficient of friction
a a constant equal to 0.03 for compression piles and 0.0225
for tension piles.
Ars the effective area ratio for the shaft.
h distance of the section of pile shaft above the pile tip
(factor allows for fatigue during installation. For long
piles the top section of the pile has large value of “h”
resulting in reduced skin fiction at that location compared
to locations further down the piles (assuming similar
values of qc).
 B. Simpson et al. / State of the art Report: Analysis and Design
cv
constant volume angle of interface friction (sand on steel),
correlated to the mean particle size (approx = 23° for D50
= 1mm rising to an upper bound of 28.8° for D50 =
0.2mm).
Ars is equal to 1.0 for closed ended piles and can be
calculated for open ended piles from the following: Ars = 1-
IFRmean (Di / D)² where IFRmean is the incremental filling ratio
and, in the absence of measured data is equal to IFRmean =
0.2 0.3
min(1,(Di /1.5) ). Values of Ars are presented in Figure 3.3
where it can be seen that a thin walled pile (Di/ D = 0.98) has
approximately 40% of the shaft friction of a closed ended piles
(Di / D = 0)
Figure 3.3 relationship between Effective Area Ratio (^0.3) and
diameter for various Di/D values.
It is noted that Equation 3.8 is appropriate for use with
driven piles. Where piles are jacked then a similar approach can
be taken except that the fatigue factor (max (2,h/D)-0.5) must be
modified. The approach presented by Jardine et al (2005)
provides more data.
In order to allow for dilatancy the reader is referred to the
original paper to allow economic design of small diameter piles.
As noted above, the dilatancy effect does not increase the shaft
resistance significantly (more than 10%) for all but small
diameter piles (less than about 0.6m external diameter).
The above method of pile design is based on the results of
load tests on a large number of relatively small diameter piles,
as is fully acknowledged by the authors of the pile design
method. The potential benefit of carrying out maintained load
prototype scale pile tests is worth investigation as suggested by
White and Bolton (2005), and should be pursued where possible
(e.g. for large off-shore wind farms where the benefits of scale
may justify the initial cost of testing).
While the above suggests that design development has
progressed significantly in recent years it is also noted that
significant research continues. This is well illustrated by a
recent research project led by the Norwegian Geotechnical
Institute on pile testing and design. The project aims to test
relatively small diameter (0.3 to 0.4m diameter) driven open
tube piles in up to 6 varying soil types (see comments on the
need to test large diameter piles above). At each site, piles will
be installed and load tested over a period of 2 years to identify
how pile capacity increases with time from construction to two
years post construction. Investigation of how loading affects
pile capacity will also be carried out. While increase of pile
capacity with time is acknowledged, the research aims to
provide greater clarity on the issue. Since pile testing is usually
carried out shortly after construction, design methods typically
predict pile capacity at 10 to 20 days after construction.
3.2.3 Onshore: Bored piles - a pressuremeter approach
Compared to offshore development, for a majority of onshore
developments there will be increased constraints in terms of
noise and vibration. The result of this is that, for urban
development at least, large diameter piles tend to be bored and
2889
cast-in-place rather then driven. Whilst onshore and offshore
pile sizes and loading conditions are different, the requirement
for overall economy is comparable albeit the manner in which
economy is achieved is not necessarily the same. For offshore
construction the cost of plant (and thereby time on site) is a very
important factor whilst for onshore conditions material usage
has a larger impact on overall cost.
When considering the technical aspects, rather than the
economic aspects of design, the soil around onshore piles
experiences disturbance in a similar manner to that of offshore
piles for similar pile types (bored, driven etc). Installation
processes modify in-situ stresses, and possibly ground strengths
due to remoulding and shearing, to such an extent that soil
mechanics alone is again rarely able to predict pile performance
accurately. Hence, in the geotechnical design of onshore piles
constructed in soil, empirical procedures dominate. An
illustrative example of this is well documented in French design
practice, now enshrined in Eurocode 7 Part 2 (BSI, 2007a),
using the Ménard Pressuremeter limit pressure to assess both
limiting shaft friction and base resistance for all main pile types
in a wide variety of soil types. Such design is useful as it is
based on a large set of pile observations and is appropriately
conservative. It is does not, however, lead to an understanding
of pile behaviour.
Published design guidance (BSI 2007a) can be supplemented
by site specific correlations to optimise pile design and to
accommodate different methods of pile installation. An
example of this is bored-base-grouted piles in dense sand
(Thanet Sand) used in East London (UK) for tall buildings and
occasionally in central London where deep infrastructure
prevents founding at shallow depth (Yeow et al, 2005). Early
design procedures were based on pile load tests (Troughton and
Platis, 1989, Chapman et al, 1999). Pile base capacity was
correlated with mean effective stress rather than the more usual
vertical effective stress. Variations in the “apparent” relative
density of the sand with depth (corrected SPT blow count
reducing with depth) along with increasing fines content with
depth could not, with confidence, be used to assess pile base
capacity variation with depth. This resulted in the need for a
more robust method of profiling pile end bearing resistance with
depth.
The Menard Pressuremeter approach (Nicholson et al 2002)
was adopted and equipment extended to allow testing to 10MPa
to allow the engineering implications of the changes in the sand
with depth to be accommodated in design (typical equipment is
limited to 5MPa or less and would not allow direct limit
pressure measurement). The accepted equation for pile base
capacity using the Ménard Pressuremeter limit pressure is:
qb = ks (Plim - )+
h0 h0
(3.9)
ks empirical factor
Plim limit pressure
h0
horizontal total stress
The value of ks is purely empirical as illustrated by the
values of 1.6 for bored piles and 3.2 for driven piles in dense
sands (BSI, 2007). This derives from differences in pile base
stiffness for loading to 10% of pile base diameter (driven piles
are stiffer then bored piles in the sandy soils). For bored-base-
grouted piles based in Thanet Sand, a value of ks of 2.5 was
found to correlate with load test results. The value of 2.5 lies
between 1.6 (conventionally bored) and 3.2 (driven) as
suggested by Nicholson et al (2002). The value is clearly linked
to the nature of base grouting and as with all empirical factors is
very much method related and cannot be taken as universally
applicable.
As with all empirically based design methods the parameter
(limit pressure in this case) is only valid if the conditions
pertaining at the time of ground investigation test are similar to
those governing the design of the foundation. If the conditions
2890 B. Simpson et al. / State of the art Report: Analysis and Design
(notably effective stress regime in the soil) vary between the
time of the ground investigation testing and the design
condition, these changes must be accommodated in the design.
Such changes result from basement excavations or changes in
groundwater table. Whilst the design approach is clearly
empirical the method of correction must be based on something
more substantive in order to predict pile capacity at the design
condition. For the case of piles in sand, the method by
Troughton and Platis (1989) corrected base resistance by the
ratio of mean effective stresses ( m') at the design condition
compared to the testing condition based on measured pile
performance. Using this approach then the corrected pile base
capacity is:
qb = ( m'design / m'GI) (ks (Plim GI - h0 GI
)+ h0 GI
) (3.10)
where subscripts “design” and “GI” indicate the working
condition of the pile and the conditions existing at the time of
the measurement of limit pressure.
Menard Pressuremeter limit pressure can similarly be used to
predict shaft capacity as described in BSI (2007a) where
empirically derived graphs are presented plotting limit pressure
against shaft friction as a function of soil type and density.
Where a reduction in effective stress occurs between
pressuremeter testing and the design condition then it is
possible, for conventionally bored piles in sands at least, to
correct the limit pressure as a function of changes in horizontal
effective stress (§ įplim = plim test 8 įıh' / ıh'). Assessment of
the initial ıh' requires a knowledge of stress history (Mayne and
Kulhawy, 1982) or by direct measurement. Change in ıh' (įıh')
may be made assuming įıh' = (Ȟғ' / 1-Ȟғ') 8 įıv', where Ȟғ' is the
drained Poisson’s ratio (Troughton and Platis, 1989). A check
on the final ratio of ıh' / įıv' is necessary such that ıh' ” 1/ Ko NC
8 ıv' where Ko NC is the normally consolidate value of Ko
(§1−sinϕ') unless pile testing shows higher limiting values (such
as ıh' ” Kp 8 ıv'). It would be reasonable to assume that such
an approach is appropriate only to a reduction in limit pressure
and not an increase. Such changes in effective stress could be
due to wide excavations or changes in groundwater level
following completion of the ground investigation.
While pile design is dominated by empiricism our
understanding of soil behaviour is more firmly rooted in science
and theory. This is ably demonstrated by Ventouras and Coop
(2009) who investigated the case of reducing pile base capacity
with depth by means of triaxial testing. The tests were carried
out at stress levels appropriate to a loaded pile bases in order to
investigate the how the variations in soil grading and particle
mineralogy could effect pile capacity. The results showed that
the increase in fines content with depth was the dominant factor
resulting in reduced pile base capacity with depth due to
reduction in the ability of the soil to dilate during shearing.
While the conclusions of the research are not such that pile
capacity can be predicted for conventional design, they do
provide better understanding on how piles behave. Such
understanding results in better and safer design.
3.2.4 Piles – Carbonate soils and chalk:
The above description of offshore driven piles and onshore
bored piles is related to silica dominated sands. Where
carbonate sands are present (often lightly cemented) experience
shows that bored piles (drilled and grouted piles) may be
preferable to driven piles. This is due to the likelihood of low
confining stresses acting on the shafts of driven piles caused by
the local collapse of the adjacent soil (installation damage) and
the ability, thereafter, of the soil to arch around the pile. Such
behaviour is well documented from the Rankin field of Western
Australia (King and Lodge, 1988) where the cost of remedying
failed driven piles came to A$300m at the time (at current
prices about US$500m excluding costs associated with loss of
revenue). The impact of such soil-structure interaction is that
drilled and grouted piles (or possibly driven and grouted piles)
are a more suitable foundation type for such soils. Design
recommendations, such as those of Randolph et al (2005)
suggest a peak shaft friction (qs (peak)) as per Equation 3.11:
qs (peak) • qc (0.02 + 0.2 e
-0.04 qc/pa
) 3.11
where pa is atmospheric pressure (100kPa),
Similar behaviour is also well documented for piles driven
into chalk (low and medium density chalk) where the limiting
design shaft friction for driven piles is often 20 to 30% of the
equivalent cast-in-place pile. Again this is due to the
contractant nature of the remoulded chalk and the ability of the
chalk mass to arch around the pile shaft (Lord et al, 2002).
For initial assessment of pile behaviour in chalk or carbonate
soils, the reader is directed to Lord et al (2002) and Randolph
(2005) respectively.
The above example highlights the need to obtain a thorough
understanding of the effects of soil mineralogy and in-situ state
when carrying out pile design. For designs being carried out at
new locations and where the soil has not supported piles or
indeed the type of pile being proposed, design must not rely
only on empirical rules developed elsewhere for different
conditions. An open mind must be maintained and the ability to
load test piles must be incorporated into the programme and
cost plan.
3.2.5 Drilled and grouted piles – rock:
Drilled and grouted piles are commonly used in near-shore
locations where rock exists below the seabed and the
construction of direct bearing foundations (gravity structures) is
not cost effective (e.g. where there are significant soft sediments
above rock head). These foundations are common for port
structures, bridge foundations and wind turbines as well as for
oil and gas industry related structures. Such foundations have
become more common in recent years as a result of
developments in piling technology allowing single piles to carry
large loads (often limited by the structural capacity of the pile
shaft). These positive developments have also coincided with
heightened awareness of the dangers of construction utilising
divers or working under pressure, as in the case of sunk
caissons, further championing the case for drilled and grouted
pile foundations. In parallel with advances in construction
capability there have also been consistent advances in design, as
presented below.
Shaft Friction: Classical design of rock-socket piles is
empirically based (Rowe and Armitage, 1987, amongst others)
and relates the unconfined compression strength of the intact
rock to the ultimate shaft friction (qs ) as follows:
qs = q ci
(3.12)
or
qs = c ci
(3.13)
qs unit shaft friction (MPa)
q
, c
and are empirical constants
ci
intact rock unconfined compression strength (UCS in
MPa)
Values of c and are typically 0.2 to 0.4 and 0.5 respectively
(Hovarth et al, 1980). The range in the value of c is related to
the roughness of the rock socket.
The parameter Įq is shown in Figure 3.4 and is obtained by
combining Equations 3.12 and 3.13, giving:
–1
Įq = c ci
(3.14)
 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 3.4 shows the range of values Įq in this case for
roughness in the range of R1 (smooth, indentations less then
1mm) to R3 (grooves 4-10mm deep, >5mm wide, spacing 50-
200mm) as defined by Rowe and Armitage (1987). The plot
demonstrates the non-linear relationship between shaft friction
and rock strength. Further data is provided by Rowe and
Armitage (1987) who suggested average values of shaft friction
constants (for relatively small diameter sockets, typically
less than 0.6m diameter) as follows:
• c is equal to 0.4 for R1 to R3 sockets and 0.55 for R4
sockets (R1 to R3 as above, R4 socket quality is defined as
grooves with depth and width greater than 10mm and
spacing between 50 and 200mm).
• is equal to 0.57 for R1 to R3 sockets and 0.61 for R4
sockets
Figure 3.4 Relationship between shaft resistance and rock UCS for
socket roughness in R1 to R3 range (after Rowe and Armitage, 1984).
Note use of qu for UCS of rock, ıci used elsewhere.
It is considered that such methods are useful for preliminary
design but must be used with caution where no comparable
experience is available (same rock, same pile size, same
construction process). The values shown above are best fit
values and have been derived from load test data. Use of such
design constants within design codes such as Eurocode 7 Part 1
(EN1997-1) could be seen to fail the requirement that the design
process returns a resistance that is “accurate or erring on the
side of caution” (this requirement is not the same as a best
estimate). The requirement results in the need to assess how a
“characteristic” skin friction should be derived. Assessment of
such characteristic values of skin friction may be obtained by
adopting cautious estimates of both ci and c. However, even
when this is done several of the Eurocode 7 National Annexes
(eg the UK national annex for Eurocode 7 Part 1: BSI, 2007b)
require that a further model factor be applied to the values to
acknowledge the uncertainty in the calculation and to arrive at
pile resistances in keeping with previous national practice and
expectation.
As noted above, the use of such empirical relationships is
only acceptable when the design constants are available from
comparable pile load tests, i.e. load tests on similar sized piles
(socket size diameter and length) formed using similar
equipment (drill bit detail and flush methodology) carried out in
the same bearing stratum. Use of extrapolated data is inherently
dangerous. Where, as is likely to be the case with most large
diameter piles, good quality comparable pile load data does not
exist then it is necessary to look to design methodologies which
are theory based, and which have been checked elsewhere
against pile load test data as required by Method 2 in section 3.1
above. Appropriate design methods should allow for variability
in ground conditions, construction techniques and pile size by
means of readily measurable parameters. These parameters are
2891
then used to populate design equations. Such an approach has
been presented by Seidel and Collingwood (2001) where the
observations made by previous researchers have found a more
rigorous home. The method combines the effects of the
following:
1. Rock mass stiffness;
2. Rock socket construction disturbance and cleanliness;
3. Socket geometry (length, diameter and roughness) (see
Figure 3.5 below);
4. Rock strength (unconfined compression strength and the
critical state angle of shearing resistance, φ'cv, of a fissure
as would be measured in a shear test)
Figure 3.5 Illustrative model of pile rock socket displacement under
axial compressive load (from Seidel and Collingwood, 2001, after
Johnston and Lam, 1989)
These attributes will be considered in turn.
1. Beyond controlling the overall settlement in the rock mass,
rock mass stiffness has a significant effect on ultimate skin
friction by providing restraint to dilation on the pile/rock
interface. Dilation occurs as a function of the socket
roughness where shear displacement on the interface results
in radial strains in the surrounding rock mass and
corresponding increases in normal stress acting on the
pile/rock interface. The constant normal stiffness value (K),
is related to the rock mass modulus (Em) and it is inversely
proportional to socket diameter as shown in Equation 3.15:
K = 2Em / [(1 + Ȟғғ).D] (3.15)
K constant normal stiffness parameter
Em rock mass modulus
Ȟғғ Poisson’s ratio
D rock socket diameter.
It is the parameter K which controls the ability of the rock to
resist dilation. Clearly, as seen in Equation 3.15, the larger the
pile diameter the smaller the value of K. This is a similar effect
as seen for driven piles in sand above where the beneficial
effects of dilation are greatest for small diameter piles.
The rock mass modulus (Em) can be measured directly (eg by
dilatometer testing), or is more usually correlated with the rock
UCS, Rock Mass Rating (RMR) or rock mass quality (US
Army, 1994)
2. The construction process used to form the pile and the
cleanliness of the pile shaft are important in assessing the
efficiency by which the pile shaft can transfer load to the
ground. High quality construction leading to a clean shaft
provides highest resistance while low quality construction
leaving smear results in low resistance. This variation in
construction quality can be represented by the factor Șc as
discussed below.
3. The rock socket geometry is critical in assessment of shaft
capacity with rough sockets resulting in increased
interaction between the pile and the surrounding rock (see
earlier comments on socket resistance above for R1-R3 and
R4 roughness from Rowe and Armitage, 1987). The socket
roughness is expressed in terms of change in radius of the
2892 B. Simpson et al. / State of the art Report: Analysis and Design
socket wall ( r). Seidel and Haberfield (1995a) present a
theoretical design development of rock sockets
incorporating a Joint Roughness Coefficient (JRC) and
fractal mathematics which better describe the effects of
scale (both of asperity height and geometry) of rock socket
roughness. Such discussion is outside the scope of this
paper but is worthy of reference for future designers.
Further data is presented in Seidel and Collingwood (2001)
where examples of roughness with rock UCS is presented as
a first pass estimate. Whilst less common in practice, direct
measurement of socket roughness with a profiling tool is
also possible.
4. The final consideration is associated with rock strength as
represented by the intact strength (UCS) and fissure strength
(friction). Constant volume fissure strength (ϕ'cv) is used to
assess the pile capacity when combined with socket
roughness and the normal effective stress acting on the
pile/rock interface as a function of in-situ stress (small
effect) and stresses mobilised due to attempted dilation at
the socket to rock interface. For high strength rocks the
limiting strength may be that of the concrete rather than the
calculated (frictional) socket shear strength.
Seidel and Collingwood (2001) took the above
considerations and combined them in a dimensionless Shaft
Resistance Coefficient (SRC) that can be used to predict shear
stress mobilised at various displacements. The SRC is defined
as:
n Δr (3.16)
SRC = η c .
1 + ν 2rs
Șc construction method reduction factor as in Table 3.1.
n modular ratio (Rock Mass Modulus / rock UCS) = Em / ıci
ǻr mean roughness height of the rock socket, measured
directly or assessed from the asperity length and mean
asperity angle
rs rock socket radius
The value of SRC can then be used to assess the appropriate
value of Įq thus allowing for rock socket adhesion to be
assessed from Equation 3.12 with allowance for cleanliness and
construction, pile diameter, rock stiffness and rock socket
roughness. Figure 3.6 shows the computed values for Įq with
SRC. Further extensions of the work allow for assessment of
mobilised resistance versus displacement to build up a full
picture of pile loading.
The above comments have addressed limiting conditions in
the rock. It is also necessary to check the strength of the
reinforced concrete pile element in axial, bending and shear
modes as well as the limiting interface strength in the concrete
adjacent to the rock socket. Such considerations require
consultation with the appropriate structural design codes.
Base resistance:
Base resistance may in rock sockets may be included in
design when it can be demonstrated that cleanliness of the base
can be achieved and where the drilling and cleaning processes
have not altered the quality of the rock immediately below the
socket base. When this is the case then rock socket base
resistance can be expressed as a function of the unconfined
compressive strength of the rock, ıci, with an equivalent bearing
capacity factor as a function of rock quality and jointing. For
massive or tightly jointed rock (clean joints) with a socket depth
at least one diameter into equivalent quality rock, Rowe and
Armitage, (1987) suggest limiting ultimate base resistance, qb,
to:
qb = 2.5 ci (3.17)
Table 3.1 Construction method reduction factors (after Seidel and
Collingwood, 2001). Note *1: Care is required to ensure that the
bentonite or polymer usage is in keeping with manufacturer’s
specification and that the product is appropriate for ground conditions
and usage for forming rock sockets.
Construction method c
Construction without drilling fluid:
• Best practice construction and high level of 1.0
construction control (e.g. socket side wall free of
smear and remoulded rock)
• Poor construction practice or low quality construction 0.3-0.6
control (e.g. smear or remoulded rock present on side
walls)
Construction under bentonite slurry *1
• Best practice construction and high level of 0.7-0.9
construction control
• Poor construction practice or low quality construction 0.3-0.6
control
Construction under polymer slurry *1
• Best practice construction and high level of 0.9-1.0
construction control
• Poor construction practice or low quality construction 0.8
control
Figure 3.6 SRC and adhesion factor (Įq) (after Seidel and
Collingwood, 2001)
Where rock socket embedment is less than 1 diameter, or
where the rock joints are not tight and clean, reduction in
capacity is required. Where rock is of lesser quality it is
necessary to accommodate the rock quality in assessment of
base resistance. For this case attention to the nature of the rock
jointing and infill is required; the reader is referred to
publications such as Rock Engineering (1994) by the US Army
Corps of Engineers where a process based on rock mass
considerations is provided. Where possible load testing should
be considered and to this end use of Osterberg cells provides a
useful tool for testing components of base and shaft resistance
independently.
3.3 Pile design – lateral loading
A literature review for papers and publications on laterally
loaded piles does not reveal the same levels of development as
seen above for axial loading of piles. Recent distillations of
design guidance for lateral loading include:
• GEO (2006): assessment of single piles in soil with recourse
to Broms (1964) and Brinch Hansen (1961);
• Turner (2006): assessment of single piles in rock with
highway structure loading, recourse to Reese (1997);
• API (2000); and
• Randolph et al (2005).
Such documents provide useful summaries of current design
practice but do not reveal significant development in design
processes. Turner (2006) and API (2000) both present analysis
methods utilising “p-y” curves to model the soil support to
 B. Simpson et al. / State of the art Report: Analysis and Design
laterally loaded piles. These p-y (force – displacement) springs
have limiting values (pur) which can be calculated according to
classical solutions such as that proposed by Brinch Hansen
(1961). For limiting values of lateral resistance, Martin and
Randolph (2006) report more recent work for deeply embedded
piles in clay soils (undrained behaviour only) while Reece
(1997) provides a calculation for ultimate lateral resistance for
piles in rock as follows:
pur = r ci
D (1 + 1.4 xr / D) • 5.2 r ci
D (3.18)
where
is a reduction factor and is equal to 0.33 for an RQD of
r
100% increasing linearly to 1.0 for an RQD of 0%. Values
of r above 0.5 should be used with caution and it is
recommended that a limit on r of 0.5 is used where
comparable load test data is unavailable.
xr is depth below top of rock
Published data for p-y curves are generally for use with
single piles; they do not accommodate behaviour of piles in
groups where there may be significant interaction between piles
as well as structural restraint at the pile head influenced by pile
cap stiffness. Hence, for pile groups recourse to p-y curves
assessed from single pile load test data is not acceptable for
detailed design and alternatives are required. Alternative
approaches include the following:
• Preliminary design can be carried out by using p-y curves to
generate single pile behaviour. The results from the single
pile model can then be used to calibrate pile group computer
codes such as PIGLET (User manual: Randolph 2003b)
with single or very widely spaced piles. Such calibration is
particularly important with layered soil where the relatively
simple input to pile group programs (linearly increasing
stiffness with depth) does not lend itself to the stepwise
varying stiffness values of layered soils. The resulting
assessment of spring data can then be used for assessment
of pile group behaviour. Upper and lower bounds of input
parameters should be obtained for the single pile to allow
pile group behaviour to be investigated.
• For detailed design, especially where there is no case
history data or where pile cap interaction with the ground is
important, use of 3D finite element (or similar) computer
programs will provide the optimum analysis. It is however
noted that offshore industry practice (e.g. API, 2000)
usually reverts to methods using p-y and t-z/q-z (vertical
stiffness) for pile group analysis, such methods are included
in commercial computer programs with varying degrees of
complexity (structural stiffness of pile cap, non-linear
springs etc). These codes do not provide full soil structure
interaction and as such require a degree of experience in
their use for large groups of closely spaced piles is required.
3.4 Developments in pile design – environmental
considerations.
What constitutes good design? Historically a good design was
one that provided the required function at minimum cost, was
constructed within the programme requirements and resulted in
minimum disruption to third parties, these issues are addressed
in Section 2 in terms of code requirements. The question is
“can we do better than this?” Recently published books and
papers on foundation design for buildings show two significant
developments that should be considered in providing a “total”
design:
• reuse of existing foundations;
• planning for new foundations to be reused; and
• incorporation of piled foundations into the building’s
heating and cooling systems.
2893
3.4.1 Foundation reuse
Reuse of foundations is a critical consideration for urban
regeneration and is well documented in Butcher et al (2006a)
and papers within the 2006 conference “Reuse of Foundations
for Urban Sites” edited by Butcher et al (2006b). Foundation
reuse has the potential to:
• reduce cost and construction programmes;
• reduce the impact on archaeological resources;
• reduce disturbance to contaminated ground; and
• avoid disturbance to existing underground services and
structures.
These benefits are in part countered by the need for:
• careful and advanced planning;
• additional testing and investigation;
• more complicated design; and
• potential increased costs in design and construction of pile
caps and transfer structures.
The main tenets for foundation reuse are presented below
and may be traced from Butcher et al (2006).
Reuse of existing foundations:
• Foundation reuse is not a decision that a designer can take
alone. The decision to reuse, while laudable, must be
agreed upon by the client, the client’s insurers and the
building control authorities (City engineer or similar). In
specific cases, other interested parties may be architects and
structural engineers (to ensure correct coordination of
superstructure to substructure and to allow space for transfer
structures), future tenants of the structure and their insurers,
the list is long. The geotechnical designer can lead this
process by documenting the cost, programme and
environmental risks and benefits of foundation reuse
thereby achieving sign-on from those parties who will be
most influential in a successful outcome. The process
adopted by the designer which results in pile reuse should
be one where reasonable skill and care is demonstrated and
this should be part of the designer’s contract. It would be
impossible for a designer to provide a fit for purpose
guarantee for pre-existing structures regardless of the
diligence of his investigation and design. As such ultimate
risk, accepting reasonable skill and care by the designer will
reside with the client and, hopefully, his insurers in the final
place.
• Foundation reuse should not detrimentally impact on the
performance of the structure being placed on it, nor should
it limit the function of the new structure. If there is
reasonable doubt relating to the performance of a
foundation when reused then these doubts should be
addressed by means of measurement or redesign. All risks
and contingency plans must be agreed by risk holders before
implementation of a construction reliant on foundation
reuse.
• When reloaded, foundations will be stiffer than on initial
loading and probably stiffer than newly built foundations.
However, when foundations are loaded beyond historic
maximum loads there will be a marked non-linearity in the
load-settlement curve when compared to newly built piles
which, while less stiff, will have a more uniform stiffness
up to working load (less non-linearity). In the case of a
foundation solution using both new and existing foundations
the difference in stiffnesses during loading should be
investigated to prevent structural over-load of piles or other
elements of the sub-structure as a result of load
redistribution.
• At the limit, the ultimate load capacity of old piles,
especially those which have been loaded for significant
periods, is seen to be larger than similarly dimensioned new
piles. This potential benefit must be balanced against non-
linear settlement impacts as described above.
2894 B. Simpson et al. / State of the art Report: Analysis and Design
• Structural performance of the reused foundations should be
fully understood. As-built records of these foundations are
valuable but must be confirmed in terms of geometry (pile
diameter, length, reinforcement quantities and depths) as
well as material strengths and durability. Where there are
no as-built records, an enhanced level of testing will be
needed. It may be that insufficient confidence can be
obtained related to the existing foundations to allow reuse.
The existing structure above foundations planned for reuse
should be checked for distress (did the foundations perform
as would be anticipated?); load intensity (what load has
been applied to the piles?); and geometry (does the pile
layout match the substructure / superstructure?). If these
investigations suggest a building that is at odds with the
anticipated foundation then further investigation will be
required.
• When foundation reuse is to be adopted the existing
structure must be sympathetic to foundation reuse. Access
to piles must be possible to check location, geometry and
materials. Can load testing be carried out using the mass of
the existing building as reaction? Does demolition cause
damage to the structure of existing piles (ground movement
causing cracking of piles in bending or tension)?
The above list provides the basic considerations for reuse of
existing foundations. The following list presents the basic
requirements that present day designers and contractors should
satisfy in order that reuse of new foundations is a viable option
in the future. Planning for reuse:
• Provision of documentation for the future is valuable and a
relatively easy task. A “close-out report” should include:
full details of ground conditions, foundation design
calculations, design and as-built layout and geometry
(dimensions and depth) and materials; full records of testing
(both pile load testing and material testing) as well as pile
installation logs and non-conformance reports (with any
mitigations recorded). The designer’s/contractor’s
programme and costs should allow both the time and
resources to complete such a report; without such financial
consideration a “close-out” report will not be provided in a
commercial world! In time the report will have a value way
in excess of the cost of production. The report should be
deposited with the building records maintained by the
owner. Other documents to be preserved include the factual
ground investigation and design reports as well as the piling
specification and contractor’s method statements.
• Where possible, monitoring of building performance is a
useful demonstration of foundation performance; this is
usually only viable for larger structures. Such monitoring
includes settlement monitoring in its simplest form but may
also include load monitoring in piles (strain measurements
usually) or pressure monitoring below rafts. The
monitoring data will validate the design and construction to
working conditions and will provide subsequent owners
with data to assess future foundation performance.
3.4.2 Energy piles
Energy piles are conventional load bearing piles which have
been equipped with ducting to carry a circulating thermal fluid,
usually attached to the reinforcement cage as illustrated in
Figure 3.7. The ducting allows the piles to be used as heat
exchange elements as well as being load carrying structures.
The thermal fluid in the ducting is usually allied to heat pumps
to amplify temperature differences. This allows the energy piles
to be used as part of the building management system (BMS) by
providing all or part of the heating / cooling demand of the
building.
The use of energy piles has been widespread in continental
Europe since the 1980’s. Elsewhere, it is gaining importance as
a means of improving the green credentials of new build
structures.
Figure 3.7 Pile with tubing installed prior to pile construction. The
tubing is often high density polythene plastic pipes of 20 or 25mm
diameter. The circulating fluid typical comprises either water, water
and an anti-freeze solution or a saline solution depending on the
operational temperature range
Brandl (2006) summarises the current considerations for
energy pile design in a detailed presentation of his 2001
Rankine Lecture, illustrating good practice by means of case
history data and analysis. Energy piles have the ability be used
cyclically by extracting heat from the ground when the building
is cooler than required and providing “coolth” when the
building is warmer than required. The “energy” part of energy
piles is usually subservient to their load carrying capacity, so
care to prevent the cyclic cooling and heating of the piles and
surrounding ground from causing damage to the soil matrix and
pile structure is a prerequisite. Brandl describes the process
where the heating-cooling cycle is annual and this is the most
typical requirement for domestic situations. It is however noted
that the heating-cooling cycle can be daily allowing
significantly higher temperature differentials to be achieved.
Such systems require more sophisticated management and are
more appropriate to commercial buildings where peak daily
conditions require cooling during day time and heating during
night time (e.g. office buildings with significant computer usage
or retail with high energy lighting). Where energy piles are
operated with conductive fluids at less than 0°C, greater care in
the design will be needed to ensure that temperatures less than
0°C are restricted to the pile structure and do not reach the
ground. In such situations accurate modelling of thermal
properties of materials and the construction geometry of a the
piles are important; Figure 3.8 shows a model simulation of an
energy pile after 15 hours operation with a circulating fluid at -
5°C. It is clear that the pile edge temperature has dropped to
1°C. Further operation at a temperature of -5°C would result in
local, increasing to general, freezing of the ground around the
pile and then potentially catastrophic thawing. The rate and
duration of energy extraction from the ground should be
analysed using thermodynamics to check that the ground does
not freeze (and detrimentally thaw at a later date) and that the
pile structure is not damaged due to temperature induced
stresses. An investigation of the latter of these issues has been
carried out by Bourne-Webb et al (2009) where thermal changes
where induced on an energy pile during (maintained load)
preliminary load testing. The pile load test demonstrated a
relatively simple model associated with skin friction on the pile
shaft being mobilised to resist both applied pile load and
thermally induced strains. The result is modified stresses in the
pile varying with depth and the need to account for these
stresses in structural design of the pile. Bourne-Webb et al also
note that the peak skin friction mobilised during the thermal test
 B. Simpson et al. / State of the art Report: Analysis and Design
was marginally lower than the expected design value. Whilst
the difference between mobilised and expected skin friction was
not particularly large (17%) it is further evidence that the design
of energy piles must consider how thermal effects will impact
on static design, especially where partial factors on skin friction
are low. More maintained load testing is required with thermal
variations superimposed on the pile.
Figure 3.8 Section of a pile model with circulating fluid at -5°C after
15 hours operation.
In general, thermal piles are usually associated with
conventionally bored cast-in-place piles, although driven
reinforced concrete piles with integrated absorber pipes are
occasionally used. CFA piling, requiring the plunging of the
thermal ducting and reinforcement cage into wet concrete, is not
generally advised due to the possibility of damage to the
thermal pipes during cage installation into wet concrete.
Nevertheless it is anticipated that development of thermal CFA
piles will continue due to the economic benefits of such piles.
Construction modifications, beyond that which is needed for
structural considerations, include stiffening the reinforcement
cage to help support and protect the thermal ducting for systems
with high energy demand and the development of plunging
mandrels to install the ducting below the cage in CFA piles
where lower energy demands are appropriate. For details on
design the reader is referred to Brandl (2006) in the first place
and thereafter to 2009 issues of Géotechnique for the
Symposium-in-Print on “Thermal behaviour of the ground” (in
preparation at time of this article being submitted).
3.5 Conclusions – Piling developments
It is hoped that the above sections show how the field of pile
design, analysis and construction continue to develop.
Examples of offshore driven piles in sand and rock socket piles
have been presented with shorter comments on axial design of
piles by means of the Menard Pressuremeter. The common
thread that these two examples exhibit, especially for shaft
friction, is that of detailed assessment of data within a
framework that is related to the model by which skin friction is
mobilised. Care must be taken with extrapolating these
methods to pile sizes beyond their respective data bases even
though both methods include pile diameter effects.
The second topic addressed in this chapter is that of
sustainable design. The need to achieve not only economic but
sustainable designs is ever pressing. Such pressures should
facilitate cross fertilisation of skills and knowledge from other
areas of engineering as well as encouraging designers and
contractors to look to the future in all that they do. The move to
include more than one function in the design of a pile is
laudable as are the benefits of reusing existing piles and
providing documentation to encourage future reuse of current
piling projects.
2895
4 EMBANKMENTS AND SLOPES
4.1 Introduction
In this section of the paper, recent developments in the analysis
and design of embankment sand slopes is described, with a
particular emphasis on Japane≤se practice. Slope stability is a
very important issue in a seismically active region, resulting in
active development of both empirical and analytical methods of
design.
4.2 Basic elements of design – embankments
4.2.1 Classification of embankments and items to be
considered in the design of cut slope
Figure 4.1 shows the main types of embankments considered in
this paper: road embankments, railway embankments, reclaimed
areas for housing lots and buildings, river dikes and sea dikes.
Design methods for dams are more severe and are not included
here. Items to be considered in the design are: fill material,
density of embankment, gradient and height of slope, and
drainage facilities. In addition, seepage of water must be
considered in river and sea dikes.
Figure 4.1: Types of Embankments
4.2.2 Fill material
Selection of appropriate fill material is important because bad
material may cause poor workability, settlement or instability.
Good fill materials have the following properties:
1. Placing and spreading, and compaction of the material is
easy.
2. Shear strength and bearing capacity are high, and
compressibility is low.
3. Water intake swelling is low.
4. They are stable against erosion and shear strength does not
decrease due to saturation.
Well graded gravelly or sandy soils satisfy these conditions
and are recommended for use as fill material. On the contrary,
expansive soils such as bentonitic or solfataric soil (sulphurous
volcanic soil) and highly organic soil cannot be used to
construct embankments. Attention is necessary to the following
special soils:
a. Volcanic cohesive soil: clay originating from volcanic ash.
Void ratio and natural water content are very high and unit
weight is very low. Therefore, trafficability is low, and the
embankment may suffer from slope failure or long-term
settlement.
b. Some sedimentary soft rocks such as mudstone, shale and
tuff: these may suffer slaking and water intake swelling
after filling.
4.2.3 Density of embankments
Appropriate placing and spreading, and adequate compaction
must be conducted during construction. In general, the
2896 B. Simpson et al. / State of the art Report: Analysis and Design

compaction control standard is decided before construction. in which u is the pore water pressure including excess pore
Several compaction control methods are available as follows: water pressure due to liquefaction, and kh: is the horizontal
a. Compaction control by density based on degree of seismic coefficient.
compaction. Fellenius’ method is used for simplicity, although its
b. Compaction control by degree of saturation. shortcomings are acknowledged (Lambe & Whitman 1979).
c. Compaction control by strength of deformation. One simple method to estimate excess pore-water pressure due
d. Compaction control by compaction method. to liquefaction (Japan Road Association, 1986) is to assume that
Ru is 1.0 when FL≤1.0, where, Ru is the ratio of excess pore-
4.2.4 Gradient and height of slope water to effective overburden pressure. When Ru = 1.0, the
In general, failure of an embankment occurs for one of two entire weight of overburden is taken on the water pressure.
main reasons: a) failure due to insufficient strength of the A new design concept to introduce performance-based
foundation ground during construction of the embankment, and design (PBD) has been developed recently. In the PBD,
b) failure triggered by heavy rains or earthquakes after deformation of an embankment must be evaluated. Several
construction. In the case of dams, their height and inclination of methods studied recently are:
slope are evaluated by conducting slope stability analyses. 1. Dynamic response analysis (FEM)
However, in case of road and railway embankments, 2. Static residual deformation analysis (FEM)
inclinations of their slopes are usually designed empirically 3. Newmark’s method (Newmark 1959)
without conducting slope stability analyses, because their length
are very long. In the empirical approach, the appropriate 4.2.5 Drainage facilities
inclination of a slope is decided by the height and the material Drainage is very important to construct safely and maintain
of the embankment. The empirical design method is not uniform stable embankments. Usually seepage water and surface water
throughout the world because fill and ground conditions are must be cut off from surrounding area into the construction site
quite different in each country. For example, road embankments temporarily, during the construction of embankments. In the
are designed based on Table 4.1 in Japan. design of permanent drainage, two forms of drainage must be
considered as shown in Figure 4.2: (a) drains for surface water,
Table 4.1 Empirical design method for road embankment in Japan including subsurface drainage and slope surface drainage, and
(Japan Road Association, 1999) (b) facilities against seepage water into the embankments, such
Fill Material Height of Inclination as horizontal drains, horizontal blankets, drainage pipes or
embankment of slope gabions.
Well graded sand, Less than 5m 1:1.5 to 1:1.8
Gravel, Gravel with 5m to 15 m 1.1.8 to 1:2.0
fines
Poor graded sand Less than 10m 1.1.8 to 1:2.0
Less than 10m 1.1.5 to 1:1.8
Rock, Muck
10m to 20m 1.1.5 to 1:1.8
Sandy soil, Hard Less than 5m 1.1.5 to 1:1.8
clayey soil, Hard clay 5m to 10m 1.1.8 to 1:2.0
Volcanic clay Less than 5m 1.1.8 to 1:2.0

If the embankment is high and/or the material is soft, low

inclination is selected. However, slope stability analysis is
necessary if conditions for the foundation ground and/or the
embankment are bad as follows:
a. the underlying ground is soft.
b. the embankment is constructed in landslide area.
c) the height of the embankment is greater than the heights
listed in Table 4.1.
d. the fill material is bad, such as high moisture clay or
volcanic ash.
e. houses exist near the embankment that could be damaged if
the embankment is deformed.
For stability, circular slip surface analysis is widely used.
However, compound slip surface analysis must be conducted if
a thin soft layer exists beneath the embankment or if the surface
of the foundation ground is inclined. Procedures to estimate
safety factor by finite element methods have also been
developed recently.
In seismic design, the seismic force is added in the stability
analysis. Moreover, the effect of excess pore water pressure is
considered if the foundation soil and/or fill material are
liquefiable. One formula considering the excess pore-water and
derived based of Fellenius’s method is as follows:
(4.1)
Fs =
¦ [cl + {(W − ub) cos α − k W sin α }tan φ ]
h
§ h · stability analysis is necessary even in normal soils.On the
¦ ¨©W sin α + r k W ¸¹
h
Figure 4.2: Drain facilities for embankments
Special care for drainage is necessary to the following
locations:
a. Small rivers or boundaries between cuts and embankments
where flows of surface water concentrate
b. Valleys, and boundaries between cuts and embankments
where much spring water emerges.
4.3 Basic elements of design- Cut slopes
4.3.1 Items to be considered in the design of cut slopes
In the design of cut slopes, slope gradient, drainage facilities,
slope protection works and maintenance works must be
evaluated.
4.3.2 Design of slope gradient
If the natural ground of the slope is composed by collapsible
soils such as shown in Table 4.2, soil investigation, soil tests
and slope stability analyses are necessary to design appropriate
slope gradients. If there is a flow of water in the ground, slope
contrary, if the slope is protected from water flow, the gradient
of the slopes composed by normal soils is often designed by
empirical methods without conducting soil investigation, soil
tests and analyses. In the empirical methods several factors,
 B. Simpson et al. / State of the art Report: Analysis and Design 2897

such as hardness of soils, height of cut slope, are considered. The planting treatment is conducted by spraying of seeds,
The empirical design method is not uniform throughout the pasting of special mats which are processed by soils and seeds,
world because soil conditions of slopes are quite different in or turf work. In the slope protection by concrete pitching etc,
each country. For example, road cut slopes are designed based four types of methods are in use:
on Table 4.3 in Japan. a. protection against weathering and erosion: guniting
(spaying mortar), shotcreting (spraying concrete), stone
Table 4.2 Collapsible soils facing,
Soil type Typical soils
b. protection against flow of soils due to erosion and spring
water: gabion
Weak for erosion Welded tuff, weathered granite
c. protection against sliding: concrete pitching, concrete crib
Unconsolidated or weathered Talus, Volcanic soil, Colluvial
work.
deposit
Weathering speed is high Mudstone, Tuff, Shale, Slate,
Serpentine
Fissured Shale, Serpentine, Granite,
Andesite
Fissures forms dip slope Slate
Sandwiching soft layers Fault clay, Post landslide or
collapsed site

Table 4.3 Empirical design method for road cut slopes in Japan (Japan
Road Association, 1999)
Soil of cut slope Height of cut Inclination of
slope cut slope
Hard 1:0.3 to 1:0.8
rock
Soft rock 1:0.5 to 1:1.2
Sand Not dense and More than 1:1.5 Figure 4.3: Drainage facilities for cut slope
poor graded
Less than 5m 1:0.8 to 1:1.0 4.3.5 Retaining walls
Dense
Sandy 5m to 10m 1:1.0 to 1:1.2 If the cut slope is steeper than the stable inclination,
soil Less than 5m 1:1.0 to 1:1.2 construction of a retaining wall at the toe of the slope is
Not dense
5m to 10m 1:1.2 to 1:1.5 appropriate. If the wall is high, such as more than 5m, stability
Dense or well Less than 10m 1:0.8 to 1:1.0 of the wall against earth pressure must be checked by
Gravelly graded 10m to 15m 1:1.0 to 1:1.2 calculation; for lower walls, empirical design based on
sand Not dense or Less than 10m 1:1.0 to 1:1.2 experience may be used.
well graded 10m to 15m 1:1.2 to 1:1.5
Clay Less than 10m 1:0.8 to 1:1.2 4.3.6 Maintenance
Gravelly Less than 5m 1:1.0 to 1:1.2 Ground and protection work of cut slopes becomes deteriorate
clay 5m to 10m 1:1.2 to 1:1.5 with age owing to several factors. The most important factor is
rainfalls and seepage from bedrock, for which appropriate
For slope stability, compound slip surface analyses or treatment is necessary based on periodical site inspection of
circular slip surface analyses are commonly used for shallow seepage water and drainage conditions.
slides and deep slides, respectively. In seismic design, seismic
force is added in the stability analysis. However, this is seldom 4.4 Basic elements of design - natural slopes
to be conducted. In the new design concept to introduce
performance-based design, deformation of cut slope may be
evaluated by dynamic response analysis or Newmark’s method. 4.4.1 Classification of landslides
But more studies are necessary to introduce these methods. There are many types of landslides of natural slopes. If the
landslides are classified by speed of sides, landslides are
4.3.3 Drainage facilities classified into two groups:
For cut slopes, several kinds of facilities must be equipped for 1. Rapid landslides
drainage as schematically shown in Figure.4.3. These drains are Steep slopes slide rapidly during heavy rainfalls or
classified into two groups: (a) facilities against surface water, earthquakes. Slipped masses fall down to the foot of slopes.
such as catch ditches at the tops of slopes, drain ditches on Surface slides of weathered soils are dominant, but if some
berms and drain channels along slopes, and (b) facilities against deep layers are week, deep slides may occur during
seepage water and spring water, such as drain conduits at the earthquakes. For example, the maximum depth of the slip
toes of slopes, horizontal drainage pipes and drainage wells. surface of Ontake Slide which occurred during the 1984
Naganoken-seibu earthquake was about 180m. Stability of
4.3.4 Slope protection works against sliding, weathering or natural slopes is generally judged roughly by several factors
erosion such as angle of slope, existence of unstable rocks, histories
Slope protection works are classified into two groups: planting of slides. Stability analyses are conducted in special cases
treatment and slope protection by concrete pitching, reinforcing only. In the analyses, stability of the slides can be evaluated
grids, sprayed concrete etc. The latter is applied only to the based on peak strength of the soils. However, prediction of
following conditions because of high cost and bad appearance. unstable slopes over wide areas during future rains and
a. The soil of the slope is too hard or acid for vegetation. earthquakes is not easy.
b. The slope is unstable against sliding. 2. Slow landslide (land creep)
c. Protection against erosion is necessary because of running Gentle slopes may creep slowly and periodically due to
surface water from rainfall or springs causing slope failure. seasonal change of water level. Sliding may occur
particularly in spring if the slope is affected by melting
2898 B. Simpson et al. / State of the art Report: Analysis and Design

snow, possibly sliding a few metres each year on weak

layers, such as weathered pumice and sandstone. Stability of
the slide can be evaluated by slope stability analyses based
on residual strength. As the slides occur periodically, slide
prone sites can be identified easily, and countermeasures
can be applied.

4.4.2 Countermeasures
Two types of countermeasures against landslides may be
Figure 4.4: Deformation of an embankment constructed by EPS method
introduced: control works and prevention works. In control
due to the weight of covered soil
works, the stability of a slope can be increased by controlling
ground water level or the shape of the slope. In prevention
4.5.3 Judgment of instability of embankments by field
works, failure of the slopes is prevented by constructing some
observation
countermeasures against sliding. Some countermeasures are
If the ground is very soft, stability of an embankment is
listed in Table 4.4.
evaluated not only before construction but also during the
Table 4.4 Countermeasures against landslides
construction of the embankment. Settlement and spreading of
the toe of the embankment are measured and may be plotted on
Type of Countermeasure methods a control diagram such as Figure 4.5.
countermeasure If the plotted point exceeds a failure (critical) line, it is
Control works Surface water drainage work Groundwater judged that the embankment is becoming unstable, and the
drainage work construction cannot be continued. An appropriate measure, such
• Ground water cut-off wall as waiting several days or constructing counterweight fill, is
• Soil removal work applied. Many control diagrams have been proposed based on
• Counterweight fill method
case studies of failed embankments.
Preventing works Retaining wall
Soil reinforcement
Ground anchorage
Pile works, Shaft work

4.5 Recent developments- embankments

4.5.1 Effective use of construction generated soils

Recently reconstruction of buildings and new construction of
underground structures have increased in urban areas.
Consequently, the availability has increased of soils from
excavations and broken brick and concrete from demolition of
buildings. These materials are difficult to dispose of because of
lack of landfill. So it becomes necessary to use these materials
for fill material or other purposes. Often, excavated soils are not
suitable for fill materials because their water content is too high
and/or they are unsuitable mixtures of coarse and fine grains.
Then special treatments are conducted as follows:
a. lowering of water content by special bags or sandwich
methods;
b. treatment of soil with air foam, fibers, cement or lime.

4.5.2 Lightweight embankments
A recently developed method to prevent large settlement due to
construction of embankments on very soft ground is to use light
weight material. EPS (Expanded polyestyrene) construction
block, lightweight treated soil with air foam (see, for example,
Tsuchida et al 2001), lightweight treated soil with EPS beads, or
fly ash may be used as the light weight material. The use of EPS
method has become especially popular recently. In this method,
EPS blocks are stacked and covered by soils. If the weight of
the soils for cover is heavy, settlement at the toe of embankment
occurs as schematically shown in Figure 4.4. Therefore
attention is necessary not only the weight of EPS but also that
of the covered soils. Floating due to buoyancy and disturbance
due to strong wind must be prevented during construction.
Moreover, EPS may be burned if there is a fire due a to traffic
accident, and it can also be damaged by petrol spillage.
Figure 4.5: Diagram to judge instability of embankments (proposed by
Matsuo and Kawamura 1977)
4.5.4 Introduction of performance-based design in seismic
design
The need to evaluate not only the safety against sliding but also
deformation of embankments was recognized after the 1995
Kobe earthquake. In river dikes, the aim is to protect from
flooding, so the critical requirement of the dikes is prevent
overflow of river water, as schematically shown in Figure
4.4(a). In road embankments, emergency vehicles must run just
after earthquakes. For example differential settlement of
approaches to bridges from embankments must be within the
appropriate value for the vehicles, as shown in Figure 4.4(b).
Thus, differential settlement must be one of the critical
conditions in design of road embankments. One more reason
why the evaluation of deformation of embankments must be
introduced in seismic design, is that the calculated safety factor
against sliding, Fs, is apt to be lower than 1.0 under the Level 2
earthquake motion, even though the ground is medium dense.
Therefore Fs cannot be used in the design under Level 2
shaking. Here, Level 2 shaking motion is defined as the
maximum shaking motion. Level 1 shaking motion is defined as
 B. Simpson et al. / State of the art Report: Analysis and Design 2899
Table 4.5: Damage level for railway embankment (Railway Technical
Institute 1999)
Deformat Damage level Settlement, S (roughly
ion level speaking)
1 No damage None
2 Slight damage S<20cm
Figure 4.6: Critical conditions for river and road embankments 3 Medium damage 20cm҇S <50cm
(restoration is available
the shaking motion which occurs two or three times during the with emergency repairs)
lifetime of a structure Recently, allowable settlements for super 4 Severe damage (long term S҈50cm
levees (Japanese River Association, 1997), railway restoration is necessary)
embankments (Railway Technical Institute, 1999) and river
levees (River Bureau, 1995) have been introduced in Japanese Table 4.6: Damage level for differential settlement between abutment
design manuals and guidelines. A “super levee” is an extended and embankment (Railway Technical Institute 1999)
area of raised ground behind the main levee, on which Deform- Damage level Differential settlement
construction may take place as shown in Figure 4.7. In the ation between abutment and
manual for super levees, allowable settlements are 50 cm for the level embankment, Sd (roughly
top of the levee and face of the back slope, and 20 cm for the speaking)
ground on the super levee. As the super levees are used for
1 No damage None
residential areas, similar safety as urban areas is necessary.
2 Slight damage Sd<10cm
Hence these allowable values were introduced in the design
3 Medium damage 10cm҇S d<20cm
manual.
(restoration is available
with emergency repairs)
4 Severe damage (long Sd҈20cm
term restoration is
necessary)

Table 4.7: Relationship between Fs and settlement (Public Works

Research Institute 1997)
Safety factor of slope, Fs Settlement (maximum)
Fsd(kh) Fsd(ǻu)
1.0<Fsd 0
0.8<Fsd<1.0 0.25H
Fsd0.8 0.6<Fsd<0.8 0.5H
Fsd<0.6 0.75H

Figure 4.7: Difference between normal and super levees

In the guideline for railways, damage levels of deformation

of embankments are classified into four grades as shown in
Table 4.5 and Table 4.6 for embankments and approaches to
bridges, respectively. If the settlement is more than 50 cm it can
be judged that long term remedial work is necessary. On the
contrary, it can be judged that the damage is slight if the Figure 4.8: Definition of allowable settlement for river dike (River
settlement is less than 20 cm. Therefore allowable settlements Bureau 2007).
under Level 1 and 2 earthquake motions were decided as 20 cm
and 50 cm, respectively as shown in Table 4.5. Recently,
seismic diagnosis of existing river dikes has been conducted in
Japan. In the diagnosis, allowable settlement is defined as
shown in Figure 4.8. It is recommended that the level of a river
crest after an earthquake must be higher than the level of the
mean monthly highest water plus wave height.
In the estimation of liquefaction-induced deformation of
structures, three types of methods are available: empirical
methods, static analyses and dynamic analyses. One empirical
method is introduced in the design manual for river dikes, as
shown in Table 4.7. In this method, settlement of a dike is
estimated by the safety factor of the slope, Fs. Two values of Fs
must be calculated: Fs (kh) which considers the seismic
coefficient, and Fs(ǻu) which considers excess pore water
pressure due to liquefaction. Then the settlement is estimated by
the lower FS. The relationship shown in Table 4.7 was derived
from the correlation between settlement of damaged dikes and Figure 4.9: Relationship between Fs and settlement ratio of river dike
Fs during several past earthquakes as shown in Figure 4.9. (Public Works Research Institute, 1997)

Figure 4.10 shows another empirical relationship between

settlement of dikes of Kiso, Nagara and Ibi Rivers during the
1944 Tohnankai earthquake, and liquefaction potential, PL at the
damaged sites. As show in this figure, the settlement increased
2900 B. Simpson et al. / State of the art Report: Analysis and Design
with the value of PL. PL is calculated by the following formula
together with Figure 4.11 (Iwasaki et al., 1978).
20 (4.2)
PL = ³ (1 − FL )(10 − 0.5 z ) dz
0
where FL is the safety factor against liquefaction, equal to the
ratio of undrained cyclic shear strength to cyclic shear stress.
The term (1-FL) is set to 0.0 if FL>1.0
Figure 4.10: Relationship between PL and settlement of river dikes
(Nakamura and Murakami 1980)
Figure 4.11: Definition of PL (Iwasaki et al 1978)
For railway embankments, a relationship among settlement,
height of embankments, density, number of cycles and
liquefaction potential, PL is prepared to estimate the settlement
of embankments as shown in Figure 4.12. This relationship was
derived from shaking table tests.
Figure 4.12: Relationship between PL and settlement ratio for railway
embankments (Sawada et al 1999)
In dynamic and static analyses, several methods have been
proposed and applied to estimate the deformation of
embankments. A technical committee organized by the Japanese
Institute of Construction examined the efficiency of these
analytical approaches. In the examination, two dynamic
methods of analysis, LIQCA and FLIP, and two static methods,
ALID and Towhata’s method, were applied to seven actual river
dikes which were damaged and non-damaged during the 1993
Hokkaidonansei-oki earthquake and 1995 Hyogoken-nambu
earthquake. LIQCA and FLIP are two-dimensional effective
stress computer codes developed by Oka et al. (1999) and Iai et
al. (1992), respectively. ALID is a simplified method using
static FEM developed by Yasuda et al. (2003) by assuming that
residual deformation would occur in liquefied ground due to the
reduction of shear modulus. Towhata’s method was developed
based on a minimum energy principle. Figure 4.13 shows the
comparison between the calculate dike settlements and the
observed settlements. Settlements estimated by the empirical
approach shown in Table 4.7 are also compared in the figure.
The predicted settlements by the analytical approaches agree
fairly well. Figure 4.14 shows the analysed deformation by
ALID at a severely settled dike. These methods were also
applied to models with countermeasures tested on a shaking
table apparatus, to demonstrate the applicability of the
analytical methods to the dikes with countermeasures.
Figure 4.13: Comparison between calculated and observed settlements
(Sasaki et al 2004)
Figure 4.14: Analysed deformation by ALID (Shiribeshi-toshibetsu
River, No.1) [8]. Height of river dike: 5.3m. Settlement: 2.3 m.
(Yasuda et al 2003)
4.5.5 Development of strengthening techniques of existing
embankments (partially quoted from Yasuda (2007)
If the surface soil of the foundation ground is very soft clay or
peat, soil improvement is necessary before construction of the
new embankment. Many improvement techniques have been
developed such as sand drains, paper drains, preloading, and
cement-mixing methods. Recently, it has become necessary to
develop soil improvement techniques for existing
embankments. In particular, it is desired to develop appropriate
countermeasures for the embankments on sandy ground which
is liquefiable due to earthquakes.
 B. Simpson et al. / State of the art Report: Analysis and Design
If the soil of an embankment liquefies, it is necessary to
prevent the flow of the liquefied soil. Lowering the water level
in the embankment by installing horizontal pipe drains may
prevent liquefaction. In contrast, if the soil under an
embankment liquefies, large settlement of the embankment
occurs due to horizontal movement of the ground below. In this
case, restricting the movement by some technique, such as the
installation of underground walls at the toes of the embankment,
can reduce the settlement. Several techniques which have been
applied to existing embankments recently are schematically
shown in Figure 4.15. The Tokaido Shinkansen Railway
(Japanese Bullet Train) runs over areas where liquefaction is
possible during future earthquakes. As liquefaction may damage
railway embankments, a sheet-pile enclosure method has been
developed to protect them as illustrated in Figure 4.15(a)
(Japanese Geotechnical Society 1998). The 1995 Kobe
earthquake caused extensive damage to the Yodogawa dike.
The deep mixing method was applied to the embankment of the
Arakawa River embankment in Tokyo, as shown in Figure
4.15(b). A loose sand layer where liquefaction was anticipated
was 3 to 6 m thick. It was planned to use the deep mixing
method for a width of 10m and a depth of 24m to stabilise the
foundation ground against external seismic forces, which are
active seismic earth pressure, passive seismic earth pressure,
excess pore water pressure and dynamic water pressure. Using
external forces, a stability analysis was conducted that included
sliding, overturning, bearing capacity of the stratum, and
circular slip failure. The detailed design method of Arakawa
Dike is explained in Yasuda (2007).
Figure 4.15: Remediation methods for existing embankments
4.5.6 Introduction of seismic design for reclaimed area for
housing lots
Though slope failures in housing lots cause severe damage to
houses and loss of lives, much reclaimed land has been
developed in urban areas without considering seismic stability.
Then appropriate procedures are needed to evaluate seismic
stability of existing reclaimed areas. Recently, this procedure
has been discussed and a guideline has been proposed by the
Kanto Branch of the Japanese Geotechnical Society (2007), see
also Yasuda (2007). In the guideline, it is recommended to
evaluate the seismic stability by the following five steps.
Step 1: Finding out of filled zone
In a reclaimed area for a housing lot, it is difficult to find the
boundary of cutting and filling by site survey, if there is no
record on reclamation work. Then the guideline recommends
finding the filled zoned by comparing old and new aerial
photos. The filled zone can also be estimated by comparing two
sets of topographical maps drawn before and after the
reclamation work. However the accuracy of the estimated zone
is likely to be lower than that measured by aerial photos.
Step 2: Field investigation to select unstable slopes
The next step is the identification of existing fill slopes
which will become unstable during earthquakes. This can be
evaluated roughly by visual inspection, based on several factors
shown in Table 4.8.
2901
Table 4.8: Factors introduced in visual inspection to evaluate seismic
stability of existing fill slope
Item Factors introduced in visual inspection
Fill Height of fill, inclination of slope, Banking
material, Type of slope protection work,
Deformation of slope, Springwater, Drainage
well
Retaining wall, Type of retaining wall, Type of weeping,
house Deformation of retaining wall, Springwater,
Deformation of foundation of house
Topography Catchment, Land use on the slope
Several evaluation methods based on these factors have been
developed in Japan, which are introduced in the guideline. For
example, a rough check sheet by visual inspection for an
existing retaining wall prepared by Yokohama City is shown in
Table 4.9.
Step 3: Soil investigations and laboratory tests
For the selected unstable slopes, detailed soil investigation
and laboratory tests are necessary to evaluate the seismic
stability of the slopes more precisely.
Step 4: Analyses on slope stability or deformation
Based on soil investigations and laboratory tests, analyses
for slope stability or deformation are conducted. In the analyses
of slope stability during earthquakes, seismic force and excess
pore water pressure must be considered as shown in Figure
4.16. Then, in the guideline, methods to evaluate excess pore
water pressure are introduced at first. One simple method to
estimate excess pore-water pressure due to liquefaction the is
same as mentioned above in 4.2.4 setting Ru to 1.0 for FL≤1.0
(Japan Road Association, 1986).
Figure 4.16: Time of the occurrence of seismic force and excess pore
water pressure
As illustrated in Figure 4.16, little excess pore water pressure
is usually induced before the peak of seismic motion (Ishihara
and Yasuda 1975). Therefore two sets of analyses (a) to
consider seismic coefficient only, and (b) to consider excess
pore water pressure only, are recommended for the evaluation
of seismic slope stability, as shown in Table 4.10.
In performance-based design, it is necessary to estimate not
only the stability but also deformation of embankments. Several
methods to estimate the deformation of embankments are
introduced in the guideline. These are:
1. Dynamic response analysis
2. Newmark’s method
3. Static residual deformation method
Step5: Selection of appropriate countermeasures
The final step is the selection of appropriate countermeasures
against slope instability during earthquakes. Three types of
countermeasures against instability of existing fill slopes are
introduced in the guideline: control works, preventing works
and slope protection works. Some of these are listed in Table
4.11. Examples of typical arrangements are shown in Figures
4.17 and 4.18.
2902 B. Simpson et al. / State of the art Report: Analysis and Design

Table 4.9: Rough check sheet by visual inspection for existing retaining wall prepared by Yokohama City.

Value
Stone wall RC wall
Classific- Item for check Normal Slightly Abnormal Normal Slightly Abnormal
ation abnormal abnormal
Drainage Drain hole 0 1.0 2.0 0 1.0 2.0
condition Ground behind wall 0 1.0 2.0 0 1.0 2.0
Percolate from wall 0 0.5 1.0 0 0.5 1.0
Inclination of the ground behind 0 1.0 2.0 0 1.0 2.0
wall
Drainage facilities of the ground 0 1.0 2.0 0 1.0 2.0
behind wall
Wall Height and inclination of wall 0 2.0 4.0 - - -
Horizontal crack of wall 0 4.0 6.5 0 3.0 5.5
Vertical or cross crack of wall 0 2.5 5.0 0 1.5 4.0
Crack at the corner of wall 0 3.0 5.5 0 2.0 4.5
Horizontal movement 0 3.5 6.0 0 2.5 5.0
Differential settlement 0 4.5 7.0 0 3.5 6.0
Opening of the corner of wall 0 4.5 7.0 0 3.5 6.0
Expansion of wall 0 5.0 8.0 - - -
Tilt of wall 0 5.5 9.0 0 4.5 8.0
Corrosion of iron bar - - - 0 5.0 8.0
Total value
Judge Total value is less than 5 Judge as less abnormal by inspection
Total value is greater than 5 and less than 9 Judge as abnormal by inspection
Total value is greater than 9 Judge as quite abnormal by inspection

Table 4.11:Countermeasures against instability of existing fill slope and

arrangement plans
Type of Countermeasure methods
countermeasure
Control works Surface water drainage work, Groundwater
drainage work, Ground water cut-off wall, Soil
removal work, Counterweight fill method
Preventing works Retaining wall, Soil reinforcement, Ground
anchorage, Pile works, Shaft work
Slope protection Turfing, Grating crib work, Gunite, Sprayed
works concrete, Gabion

4.6 Recent developments - cut slopes

Figure 4.17: Countermeasures by horizontal drainage drilling and
drainage wells (Kanto Branch, JGS, 2007)
Figure 4.18: Countermeasures by ground anchors and pile work (Kanto
Branch, JGS, 2007)
During heavy rainfalls it is necessary to control the traffic of
cars or trains. Rain gauges are effective for judging the critical
condition to shut down the traffic. As a result, many rain gauges
have been installed and many methods to judge critical
conditions have been proposed. In the proposed methods
several factors are considered (see, for example, ATC3 1997) :
1. Total rainfall (for example, two weeks up to the previous
day)
2. Power of rainfall
3. Rainfall in one hour
4 Effective rainfall
5. Tank model method
Figure 4.19 shows one diagram to judge critical conditions
by continuous rainfall and rainfall in one hour. Boundaries
between failed and safe cut slopes and embankments estimated
based on case histories are shown by curves. The limit to shut
down the Japanese railways is shown by step lines.

Table 4.10: Two factors to be considered in stability analysis. 4.7 Examples of failures- embankments

Factors to be considered
Type of filled soil Excess pore Seismic 4.7.1 Slope failures and slumps of expressway embankments
water pressure coefficient during the 2004 Niigataken-chuetsu earthquake in
Loose and yes No Japan (partially quoted from Yasuda et al. 2008)
saturated On October 23 in 2004, the Niigataken-chuetsu earthquake, of
Sand
Dense and/or No yes Magnitude 6.8, occurred and caused serious damage to many
unsaturated structures and slopes in Japan. Six expressways were closed due
Clay no yes
 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 4.19: Diagram to judge critical conditions of failure of cut slopes
and embankments for Japanese railways
to the earthquake. The total length of the closed expressways
was 580 km. Emergency treatments were applied to the
damaged expressway embankments by filling, placing and
spreading. Then all expressways were able to opened for
emergency vehicles about 19 hours after the earthquake because
no serious damage was induced in expressway bridges and
tunnels. About 13 days after the earthquake all expressways
were opened for all vehicles.
Among the affected six expressways, the following two
zones were severely damaged.
1. Between Muikamachi IC and Nagaoka IC of Kan-etsu
Expressway (57.6 km) as shown in Figure 4.20, and
2. Between Kashiwazaki IC and Sanjyo-Tsubame IC of
Hokuriku Expressway (50.3 km)
Most serious damage occurred in the following sections:
a. Kan-etsu Expressway: between Horinouchi IC and
Echigokawaguchi IC (8.8 km), and between
Yamamotoyama Tunnel and Yamaya PA (5.5 km ) ,
b. Hokuriku Expressway : between Ohzumi PA and Ngaoka
JCT (6.0 km)
The section between Horinouchi IC and Echigokawaguchi
IC of Kan-etsu Expressway was constructed on the gentle
slopes of hills. In contrast, the section between Yamamotoyama
Tunnel and Yamaya PA was constructed on flat ground. In the
former section, embankments were constructed mainly by “half-
bank and half-cut” methods on the slope of the hills. Sliding of
the filled embankments occurred at several sites during the
earthquake. Where embankments were constructed by filling
soils on level grounds in the latter section, large settlements of
the embankments occurred.
Figure 4.20: Route map of Kan-etsu Expressway and Hokuriku
Expressway
2903
Several seismic records were obtained in these severely
damaged zones. The recorded maximum surface accelerations
were 0.500g at Horinouchi Town, 1.757g at Kawaguchi Station,
1.532g at k-net Ojiya site and 1.029g at Ojiya Castle. Therefore
it can be said that seismic motion in the severely damaged zones
was very strong as the maximum surface acceleration was about
0.5g to 1.7g.
A section between Koide IC and Ojiya IC of Kan-etsu
Expressway was selected to study the influence of type of
embankments upon the damage. In this section, embankments
were constructed by three methods: filling on a level ground,
widening, and half-bank and half-cut. Percentages of the lengths
constructed by these methods are 57 %, 5 % and 7 %,
respectively. The other 31 % are cuttings, tunnels and bridges.
Total lengths of damaged and undamaged embankments
constructed by the three methods are compared in Figure 4.21.
Lengths of damaged embankments seem to be two to three
times the length of intact embankments regardless of
construction method, and many sites were damaged even where
the ground is flat.
In Japan, damage to road embankments is classified in three
levels as shown in Figure 4.22, and this was applied to the Kan-
etsu expressway. Serious damage occurred at half-bank and
half-cut sections only. In the embankments on level ground,
medium or minor damage dominated. According to the
mechanism of failure, the damage of the Kan-etsu expressway
embankments in the section between Koide IC and Ojiya IC,
can be classified to three types as follows:
Type 1: Serious slide of the embankment on the sloping
ground as schematically shown in Figure 4.23(a)
Type 2: Settlement of the embankment on the level ground
without the deformation of the ground as schematically shown
in Figure 4.22(b)
Type3: Settlement of the embankment and the culvert on the
level ground with the deformation of the ground as
schematically shown in Figure 4.223c)
Locations where these types of failures occurred are shown
on Figure 4.20.
Figure 4.21: Total length of damaged and not damaged embankments
constructed by three methods
In Japan, the current seismic design method for road, railway
and river embankments is the seismic coefficient method based
on circular slip surface analysis. However, it has become
necessary to introduce performance-based design, in which it is
necessary to evaluate the performance by the deformation of
embankments, such as settlement. However, analytical methods
to evaluate the deformation during earthquakes have not been
fully developed. So, based on these case histories for road,
railway and river embankments, new design methods have been
discussed by several technical committees organized in the
Japanese Geotechnical Society or other associations. In a few
2904 B. Simpson et al. / State of the art Report: Analysis and Design

years, new seismic design methods will be introduced in design
codes for road, railway and river embankments. In the new
design codes, the following analytical methods will be
introduced:
Type 1 – Embankment on sloping ground: Newmark’s method.
Type 2 – Embankment on level ground: Dynamic response
analysis or static residual deformation analysis.
soil liquefied during the earthquake and caused settlement and
tilting of houses.
In 2003, 35 years after the 1968 Tokachi-oki earthquake, a
new Tokachi-oki earthquake occurred and caused damage to
houses in this district again. Figure 4.25 compares damaged
houses during the two earthquakes. In both cases, damaged
houses were located on fill ground. However, houses damaged
during the 1968 earthquake survived during the 2003
earthquake. After the 1968 earthquake underground conduits
were constructed in the damaged zone, probably lowering the
ground water level. This may be one of the reasons why the
damaged houses survived during the 2003 earthquake.

Figure 4.22: Classification of damage to road embankments

Figure 4.24: Soil cross section and damaged houses during the 1968
Tokachi-oki earthquake

At Utsukushigaoka district near Kiyota district, several

Figure 4.23: Classification of the damage to the embankment of
Kan-etsu Expressway according to the mechanism of failure.
Type 1: Serious slide of the embankment on the sloping ground;
Type 2: Settlement of the embankment on the level ground without the
deformation of the ground;
Type3: Settlement of the embankment and the culvert on the level
ground with the deformation of the ground.
houses settled as shown in Figure 4.26 during the 2003
Tokachi-oki earthquake. It can be judged that liquefaction
occurred in this site because traces of sand volcanoes were
observed around the damaged houses. The boiled soil was
volcanic sandy silt. The damaged ground had been constructed
recently by filling a channel in a hill zone. Therefore, it is
considered that the filled soil liquefied and caused the
settlement of the houses. Yasuda et al. (2004) measured the
angle of inclination of the damaged houses roughly, as indicated
on Figure 4.27. As shown in this figure, about seven houses
settled and tilted more than 1°. According to the experience of
the damage during the 2000 Tottoriken-seibu earthquake,
inhabitants felt giddy and nauseous, and after the earthquake
they could not live in the houses which tilted more than 1/100,
as indicated in Figure 4.26

4.7.2 Slope failures of a reclaimed area for housing lots

1. Liquefaction-induced settlements of houses on filled
grounds
An artificially filled housing lot at Kiyota district in Sapporo
City was severely damaged during the 1968 Tokachi-oki
earthquake. This might be the first experience of the damage to
artificially filled housing lots due to earthquakes, in Japan. The
maximum surface acceleration at Kiyota was estimated as about
0.08 g. The housing lot was constructed by cutting soils from
hills and filling valleys as shown in Figure 4.24. There were
279 houses in the housing lot. Of them, 27.5% of houses were
damaged during the earthquake. As shown in Figure 4.24 no
house constructed on the cut ground was damaged. On the
contrary, 56% of houses constructed on the filled ground settled
and tilted. Ground water table in the fill was shallow and filled Figure 4.25: Comparison of damaged houses during two earthquakes
soil is volcanic ash sand. Therefore it is considered that the fill
 B. Simpson et al. / State of the art Report: Analysis and Design 2905

Figure 4.26: Settled houses due to liquefaction during the 2003 Figure 4.28: Collapsed houses due to slope failure at Midorigaoka in
Tokachi-oki earthquake at Utsukushigaoka district Sendai City during the 1978 Miyagiken-oki earthquake

Figure 4.29: Procedure of reclamation work conducted at Midorigaoka

Figure 4.30: Soil cross section along failed slope (Tohoku Branch,
JSCE)

Figure 4.27: Rough measurement of inclination of settled houses

2. Collapse of houses due to failure of fill slopes

Several housing lots were damaged during the 1978
Miyagiken-oki earthquake in Sendai City in Japan, some of
them affected by a severe slope failure of an embankment at
Midorigaoka Ichi-chome housing lot. As shown in Figure 4.28,
many houses collapsed due to the slope failure. In 1957 to 1958,
the embankment was constructed by filling a valley. During the
reclamation work, trees on surrounding hills were felled and
thrown down into the valley. Rocks cut from the hill also were
thrown into the valley, then soil cut from the hills was placed as
schematically shown in Figure 4.29. Figure 4.230 shows a soil
cross-section along the slope. The thickness of the fill was 5 to
15 m. The density of the fill was very low with 0 to 10 of SPT
N-values. It is estimated that buried trees became humus within Figure 4.31: Failed slope and damaged houses at Midorigakoka in
Kushiro City during the 1993 Kushiro-oki earthquake
30 years. Moreover, as big voids existed between buried rocks,
groundwater flowed in the big voids and caused weathering of
At Kayanuma district in Shibecha Town, land for housing
rocks. Then the bottom of the fill became very loose and caused
lots had been developed by cutting the north slopes and filling
the slope failure during the Miyagiken-oki earthquake.
in the south zones as shown in Figure 4.34. Four valleys were
During restoration, steel pipe piles, 318.5 mm in diameter,
filled with a maximum thickness of 10 m. No special drainage
were installed to increase resistance against sliding, as shown in
conduits were put on the bottom of the valleys. Filled soils slid
Figure 4.30. The steel pipes was strengthened by filling with
and nine houses were severely damaged during the 1993
steel H sections and concrete. The steel pipe piles were installed
Kushiro-oki earthquake as shown in Figure 4.35. Soil
in two rows at 2m centres. A concrete retaining wall and
investigations by Swedish Weight Sounding and other tests
drainage wells were also constructed. The safety factor against
were carried out after the earthquake to study the mechanism of
sliding during anticipated earthquakes was thereby increased to
damage. The fill soil was volcanic silty sand. Figure 4.36 shows
1.2.
a soil cross section along lines A-A’ and B-B’ which cross
heavily damaged zones. The filled soil was loose with SPT N-
2906 B. Simpson et al. / State of the art Report: Analysis and Design
vales of around 10, and the lower part of the fill was saturated.
It was difficult to recognize the occurrence of liquefaction
during the site survey conducted just a few days after the
earthquake, because the land was covered with snow and sand
volcanoes could not be observed. However, detailed seismic and
liquefaction analyses suggested the occurrence of liquefaction
of filled sand. In the slope stability analysis, the safety factor
against sliding was less than 1.0 if the seismic acceleration
exceeds 0.2g Unfortunately the actual acceleration is not known
because no seismic records were obtained around here.
Deformation analysis by Tara 3 showed large deformation of
about 1m, as shown in Figure 4.37. Deformation of the area
could be estimated by FEM as shown in this figure because the
slide was not large.
Figure 4.32: Location of damaged houses at Midorigaoka
(JSSMFE,1994)
Figure 4.33: Mechanism of slope failure (JSSMFE, 1994)
Figure 4.34: Filled zones and damaged houses at Kayanuma (JSSMFE,
1994)
Figure 4.35: Deformed bank and damaged houses at Kayanuma during
the 1993 Kushiro-oki earthquake
Figure 4.36: Soil cross section along the slope (JSSMFE, 1994)
Figure 4.37: Deformation of bank analysed by Tara 3 (JSSMFE, 1994)
5 UNDERGROUND CONSTRUCTION
5.1 Introduction
This section of the paper addresses design and construction of
underground structures, principally tunnels, and presents
observations and findings based on experience gained over the
past two decades during the design and construction of major
railway, highway and water-transfer tunnel projects in the
United States, Africa and Asia. During this time it has been
possible to compare the performance of the as-constructed
structures to the predicted performance as given by the design
process. Excavation was typically carried out in widely varying
ground conditions (weak-bedded rock, jointed hard rock, and
soil-like poorly cemented rock with groundwater). It is therefore
intended that the design guidelines could be applied, in general
terms at least, to any tunnel excavated in any ground mass
condition with any ground material type (hard rock, soft rock,
bedded rock, soft ground, and soil) with or without ground
water.
It may be queried as to why there is a basic review of tunnel
design in a state-of-the-art paper. The purpose of this
rudimentary review is to have designers and other design team
members, inclusive of managers, realize that even with the
advances of numerical methods and the tools available to assist
in technical evaluations, it is still essential not to lose sight of
the fundamentals. Common sense and a basic understanding of
the fundamental physics involved, along with comprehensive
checks during both design and construction are still imperative,
even in the presence of all the computer wizardry.
 B. Simpson et al. / State of the art Report: Analysis and Design
The paper provides a basis for design of underground
structures, specifically tunnels, by separately addressing the
following: Basis of Design, Design Methodology, Design Codes
and Analysis, Materials and Sustainability, Performance
Monitoring and Construction Reviews and Future
Developments.
Increasingly, concern for the environment and spiralling
costs for commodities such as steel, concrete and fuel have also
driven owners to consider and introduce specific requirements
for sustainability. So one intention of this paper is to
demonstrate how these sometimes conflicting requirements can
be addressed through innovation and quality in the design,
procurement and construction process.
5.2 Basis for design
5.2.1 Definition of design
The basics of tunnel design and achieving the functional
requirements through the design of large span soft ground
tunnels based on the sequential excavation and support method
(SES) or Cut-Cover tunnels on a design-build or design-bid-
build contract delivery project requires a well planned process
that defines the objectives of the design as well as the process to
follow while executing the design. This first essential
requirement is usually addressed through the contract
documents which govern the administration of any given
project.
As more projects follow the trend of transferring project
design and construction to a single entity and design service life
increases to 80, 100 or 120 years, it becomes more important
that contract documents clearly state the requirements expected
of the design. Continuity in design process becomes paramount
in ensuring that the finished product meets the expectations of
the Owner. This is clearly evident for design-build contracts,
but often the traditional design-bid-build projects also suffer
from a lack of proper understanding of the development of
tunnel design.
A tunnel design typically consists of six major attributes
which define the required work:
• Each tunnel design is individual
• Each tunnel design requires the excavation method to be
selected
• Each tunnel requires the decision to be either Drained or
Undrained which defines the type of loading the lining must
withstand
• Each tunnel design should define the standards and
analytical tools to be used
• Each tunnel design typically has a unique set of “Owner”
requirements
• Each tunnel design is not complete until the excavation is
completed and the correct lining installed.
It may be asked, what is design? It may be thought of as any
rational/verifiable/justifiable process whereby a concept is
transformed into information that can be used to bring the
concept to reality. In the case of a civil structure, design may be
defined as that process, which is documented and can be
audited, that determines information such as the shape (form) of
the structure, the dimensions (size) of the structure, the
thickness and strength of structural members that comprise the
structure, etc. This information can be used to compute the
time, money and resources required for construction of the
structure. An optimum design may be thought of as one that not
only maximizes the functionality and safety of the structure and
minimizes its cost, both from an initial (capital) cost and a long-
term (maintenance) point of view, but also takes advantage of
known sustainability options available as well.
A very important requirement of a sound design is that there
exists documentation of the process that was used, so that the
design work can be examined later in relation to the measured
or observed performance of the structure. Such an audit review
2907
would also be required for contractual, insurance purposes, etc,
in the event of unsatisfactory performance of the structure.
During the design process typical Design Management
Elements include but are not limited to:
• Collect data and develop concept
• Project hazard analysis
• Defining design requirements
• Define design protocol of codes and standards.
• Define quality processes
• Technical risk management
• Technical studies
• Interface management
• Technical review
• Environmental management
• Internal value engineering
• Producing designs and documents
• Clarifications
• Correspondence, mtgs. and communications
• Change identification and implementation
Careful management of these issues will assist in the
development of an appropriately documented integrated design
as is discussed later in this section.
5.2.2 Types of underground structures
For the purpose of this paper, a tunnel is regarded as an
underground opening of limited span (width) and height
(generally less than about 15 m), oriented in a horizontal/sub-
horizontal direction, which is advanced linearly by rapid
excavation techniques generally through a variety of geological
materials. This section will not specifically cover other
underground excavations such as shafts (which are oriented in a
vertical/sub-vertical direction) and caverns, which have
significant span and height and are more three-dimensional in
nature.
Cut-and-cover construction, as the name implies, involves
excavating a trench (cut), constructing the tunnel structure, and
backfilling to restore the surface to its original condition
(cover).
The concept of cut-and-cover construction dates back to at
least 2180BC, when Babylonians constructed a brick arched
tunnel, approximately 12 ft. high and 15 ft. wide in a trench
across the Euphrates River as noted in the Questia Online
Encyclopaedia (www.questia.com/library/encyclopedia/tunnel.jsp).
The river flow was first diverted prior to construction and
reinstated after the tunnel was backfilled.
With its inevitable surface disruption, cut-and-cover
construction in highly trafficked urban environments can be
intrusive. However, for a number of reasons – shallow
alignment, large and/or varying cross section, prevailing ground
conditions – which may make conventional tunnelling risky
and/or uneconomic, cut-and-cover construction remains a viable
and popular alternative.
Bored tunnels are typically employed when costs of cut-
cover tunnels exceed that of mining techniques or surface
disruptions are not tolerable. Tunnel bores may be advanced by
a tunnel boring machine (TBM), generally with the use of a
segmental concrete or iron lining, or by other techniques.
Approaches which do not use a TBM include SES –
sequential excavation and support method, also known as SEM
– sequential excavation method. This form of excavation
requires providing support in the form of ribs and lagging or
shotcrete and so is also known as sprayed concrete construction.
In the highly stressed rocks of the European Alps, this approach
has been developed as the NATM – New Austrian Tunnelling
Method – in which support is provided to the opening based on
the observed behaviour of the ground while adjusting to the
revised stress fields.
5.2.3 Performance requirements
Owners’ performance requirements for underground structures
have traditionally included safety, economy and durability for
2908 B. Simpson et al. / State of the art Report: Analysis and Design
both construction and subsequent operation over and beyond the
specified design life of the structure.
To assist with a common understanding of design terms, the
following definitions are suggested. However any given project
often has a set of predefined conditions to which the design
must adhere. It is suggested that when this occurs the project
team clearly define the terms and how they are intended to be
used in the design of the structure. Following this suggestion
will help to eliminate any misunderstanding between Owners,
Designers and Contractors.
Tunnels must be safe as society demands security in the
structures it inhabits or uses – thus tunnels are often classified
into two categories:
• Inhabited space
• Uninhabited space
These definitions help clarify the level of risk that needs to
be managed and where that risk may reside, i.e., life safety or
project capital costs.
Having defined the intended use of the underground space, it
is imperative to understand the requirements for which the
structure is required to perform. The definitions of the types of
analyses to be incorporated in the design provides a sound basis
for guiding the design process. The definitions below serve well
to guide the designer in this effort and are closely related to the
functional requirements that are discussed below.
ULS – Ultimate Limit State Design may be viewed such that
exceeding the limit results in failure. It is irreversible and may
result in risk to loss of life and/or structure
SLS – Serviceability Limit State Design may be viewed as a
hindrance or undesirable performance, but this may be
reversible and does not risk loss of life or structure.
5.2.4 Functionality requirements
The type of tunnel to be designed and constructed often defines
the requirements. Generally there are five types of tunnels and
each category generally has its own specific requirements for
functionality. The most common types of civil works tunnels
are:
• Road Tunnels
• Railway Tunnels (Mass Transit, Commuter, Freight, High
Speed)
• Hydraulic Tunnels (Water, Wastewater, Storm Water,
Head/Tail Race)
• Utility Tunnels
• Pedestrian Tunnels
Each of the above tunnels generally requires a clear
statement of performance requirements regarding but not
limited to the following:
• Tunnel cross-section(s) (twin bores versus single bores)
• Design life
• Tunnel orientation
• Durability
• Alignment (horizontal and vertical)
• Lighting
• Portal locations
• Power
• Structure gauge/size
• Aerodynamics
• Water tightness /leakage
• Noise
• Ventilation/purging
• Vibration
• Drainage
Other issues which generally require early decisions include:
• Fife safety issues and requirements
• Security requirements
• Orientation of tunnel with regards to geological setting /
design requirements
• Drained or undrained tunnel design (usually
environmentally driven)
• Serviceability requirements (allowable
displacement/deformation/cracking)
• Environmental issues / handling of tunnel spoil /
groundwater pumping
• Project cost restraints / budgets / schedule
However, even with properly defined Owner Requirements
there must always be a clear traceable path of evidence that
shows that the client requirements for function, operation,
finished size and shape have been achieved and that life-safety
issues have been identified and duly addressed.
The Role of Geotechnics
An important difference between tunnel design and any other
civil construction is the percentage that geotechnical parameters
influence the design and construction and are quantifiable in
comparison to the other design data. For example, in the design
of a bridge or building, as illustrated in Figure 5.1, the
geotechnical variables influence only the foundation, which is a
minor part of the construction, and so give rise to little financial
risk during construction. Whereas in tunnel design, these
parameters influence the excavation and support which is the
greater part of the design and construction; this results in a high
risk on a major part of the financial cost associated with the
work.
Hence the age old saying:
“Geotechnical Investigations?? ….. Pay Now or Pay Later”.
Accordingly, the value of good geotechnical data and
applied geotechnical engineering is essential to a successful
tunnel project.
Figure 5.1 Role of geotechnical influence on design
5.2.5 Site investigations
A tunnel is a structure that is typically controlled
geotechnically. The predominant loads acting on the structure
are the ground pressure load and, if groundwater is present, the
groundwater pressure. Other loads include dead load associated
with the liner and any non-removable elements and the live
loads as well as surcharge loads.
A tunnel design is only as good as the geological and
geotechnical data. An important step in the design process is the
determination of a “best estimate” of the anticipated geological
and geotechnical conditions along the planned tunnel alignment.
This information should include the following:
• Ground types (e.g. sandstone, silt, clay, sand, etc)
• Geological structure (bedding, jointing, faults, shear zones,
etc)
• Ground mass strength and stiffness
• Groundwater characteristics (location of water table(s),
permeability, etc).
An adequate site investigation program prior to construction
should always be carried out for this purpose. The work must
be sufficient to identify any adverse conditions which might
 B. Simpson et al. / State of the art Report: Analysis and Design
occur along the tunnel alignment. Inadequate data often leads
to excavation delays, slower progress, which increase the time
required, and thus the cost, to excavate, support and line the
tunnel.
It is often not possible to determine the exact distribution of
conditions along the tunnel drive, so often “representative”
conditions are established. These then serve as “benchmark”
conditions that can be later compared to the actual conditions
encountered. Historically, for any given region where
geological conditions are equivalent and reasonably understood,
it is possible to arrive at an estimated percentage distribution of
ground types to be assumed for use in the “base” design for
bidding. The actual design will require verification of ground
conditions during construction with appropriate payment terms.
Once a prediction of representative geological and
geotechnical conditions along the planned tunnel drive has been
achieved, the next step is “geotechnical design”. This is the
process of predicting the response of the ground mass in each
representative condition along the tunnel to the planned
excavation of the tunnel. This prediction is required to assess
what construction method, excavation method, support
measures and ground treatments may be required to achieve
stabilization of the excavation. For small span tunnels in “good”
ground minimal, if any, measures are required and routine
construction and excavation methods are possible. In “bad”
ground, however, particularly where the tunnel is large in size,
extensive complex procedures may be required.
5.2.6 The role of water and water pressure
Of particular note, the designation of a drained or undrained
tunnel is not a definition of water tightness but it is a clear
definition of the intended load which the tunnel shall be
designed to resist. A clear understanding of this design
requirement is essential to avoid catastrophic errors in the
selection of the excavation opening size and the loading which
the permanent liner shall be designed to resist. Undrained
tunnels are typically designed to resist the full water pressure as
defined by the designated groundwater table and geo-
hydrological model. Whereas drained tunnels are often designed
to resist only some portion (usually minimal/residual) of the
pre-existing hydrostatic pressure.
The decision to build an undrained or drained tunnel should
be based on actual requirements. Environmental issues should
be the most significant factor contributing to the selection of an
undrained tunnel. For example, it is understood that the German
Rail preference has recently followed this policy of making this
selection rather than the previous requirement for design and
construction of only Undrained tunnels to reduce structure
maintenance.
Often the Designers are permitted to propose design and
construction of drained tunnels where they consider the design
appropriate unless they are specifically prohibited by Contract.
The following conditions typically need to be satisfied to allow
for the design and construction of a drained tunnel:
• The reduction of the water table will not have adverse
environmental effects (such as settlement of structures,
significantly reduced stream flows, etc.).
• The inflows would not be so high as to cause maintenance
problems.
• The ground water is not aggressive (<1.5 g/l of sulphates).
Drained tunnels are typically required to resist some amount
of residual hydrostatic pressure. Often this pressure envelop
consists of an allowance for a residual radial water head of 5m
above the inner contour line of the inner lining. However, a
proper understanding of the hydrogeological characteristics of
the geology where the tunnel is driven is required. In some
cases the residual radial pressure envelop approach is
appropriate. In other cases a more conservative linearly
increasing hydrostatic pressure loading may be more
appropriate. The tunnel arch or vault area usually is not
significantly affected by this amount of hydrostatic load
2909
regardless of the type of hydrostatic loading envelop applied.
However, for non-circular tunnels, the tunnel invert will require
significant reinforcement depending on the type of hydrostatic
envelop applied and the geometry of the tunnel section.
Drained tunnels are typically required to have longitudinal
groundwater drain pipes along each side of the tunnel with
maintenance access not to exceed 50 m intervals. Where drained
tunnels are constructed the drains typically consist of perforated
pipe with either a gravel pack or no fines concrete for collection
of the groundwater. The pipes are located near, but always
below, the longitudinal construction joint at the interface
between the crown lining and the invert ‘corner’. The inner
lining is tapered above and below to provide space for the
drainage collection pipes.
Undrained tunnels are subjected to the full hydrostatic load
as defined by the design groundwater level for the project or
section thereof. Some projects establish more than one design
groundwater level. In all cases save transient load cases,
appropriate load factors should be applied to the groundwater
pressures when designing the tunnel unless a “working stress”
analysis is performed. Differing design procedures apply the
load factors either to water pressure data in the analysis or to
resulting stresses in the linings. Further debate of the merits of
these alternatives will be valuable, involving both tunnelling
and geotechnical disciplines. A very applicable paper by
Bilfinger (2005) provides an approach to designing for
groundwater loads on tunnel linings.
It is also relevant to note that tunnels do not have to be dry in
order to achieve an undrained pressure loading. The only
condition that is required to attain full hydrostatic load to the
tunnel lining is a low relative permeability of the tunnel lining
relative to that of the ground surrounding the tunnel.
5.2.7 Waterproofing
The Owner’s requirements will usually dictate the required
“dryness” of a tunnel. Water tightness usually has a direct
impact on operation and maintenance costs for the Owner if the
underground space is considered as inhabited. Accordingly,
water tightness should be specified by an allowable leakage
rate. It is rarely observed that a tunnel is truly dry. Hence there
should be an allowance for acceptable leakage. This allowance
requires an easy but reasonably discernable method to
determine contract conformance.
During the course of a recent project it became clear that the
contract documents were incomplete in addressing the
requirements for water tightness for a rail tunnel where the
contract merely stated that the tunnel must be “dry”. Often
complete dryness is not required to meet the functional
requirements of the structure or that of the Owner. Accordingly
the subject was duly reviewed and further clarified to provide a
practical measure to achieve the Owner’s functional
requirement. To meet the intent of the project Design
Specifications which stated that the tunnels shall be constructed
to be completely “dry”, it has been recommended that the use of
the German Code DS 853 Class 3 as the objective criterion for
evaluating compliance with this requirement. Unless very
stringent watertightness criteria are required as for underground
system control facilities, etc., this class of “watertightness” is
considered adequate for most inhabited underground space as
typically required for transit projects.
Typical water proofing systems consist of geofabrics,
synthetic geomembranes with a series of water stops to
compartmentalize segments of the tunnel structure and post
grouting tubes to seal off leakage in the future. Designs
typically include a waterproof membrane (typically 2.5 mm
pvc) over the crown. A geotextile ‘fleece’ is usually provided
outside the membrane and external waterstops cover the full
perimeter of all transverse construction joints. Isolating
waterstops are generally required at bulkhead locations to
prevent water leakage from extending beyond the space
between the bulkheads.
2910 B. Simpson et al. / State of the art Report: Analysis and Design
Where undrained tunnels are required or proposed,
Designers typically provide a full perimeter waterproof
membrane (typically 2.5 mm pvc) to create a fully “tanked”
structure.
5.3 Design methodology
As with all project designs, a clear understanding of the design
methodology must exist which is geared to addressing all the
Owner’s requirements and those issues not directly addressed
by the Owner but which are the responsibility of the Designer to
deliver a durable and safe system as defined by the Designer’s
profession.
5.3.1 Integrated tunnel design process
Tunnels must be economical firstly with regard to cost to design
and construct and secondly with regard to maintenance. Most
design decisions are, implicitly or explicitly, economic
decisions. The life-cycle evaluation of the structure accounting
for all costs and benefits arising from all the phases of the life of
the tunnel must be taken into account. These phases include
design, construction, operation and closure or transformation
(salvage) of the facility. This concept is rapidly being advanced
in the European countries under the designation DARTS
(Rostam & Høj 2004) which is an acronym for “Durable and
Reliable Tunnel Structures”.
To account properly for the life cycle costs of the structure,
an integrated design should account for all effects on the Owner
and society over the complete lifetime of the structure as an
integrated process. The level of detail requires adjustment to
match the current phase of the planning and design and
subsequent decisions, which affect the final structure and are
required at various stages of the planning and design process.
This approach requires that all parties involved with the project
cooperate as a team and not as contractual opponents. An
adversarial relationship during a project is a recipe for
marginalizing the design, and risks missing the optimization of
the project. The product, as seen by the Owner and Designer,
will influence the design, specification, contracts, and the
organization and structure between the various parties during
execution and operation. Obviously this affects how the
interfaces between the Owner, Designer, Contractor and other
third parties are selected and administered over the life of the
project to result in an integrated design. This is important as
most tunnels are designed and constructed to serve the public
and the complete life-cycle of the project should be optimized to
their benefit
During the process of modelling and selecting design options
and solutions, there is always uncertainty in the quantification
each issue. All potential outcomes of the design process must be
considered, evaluated and documented. Omission of any
potential outcome may incorrectly favour solutions which do
not provide real benefit.
Uncertainty may not be limited to the physical or statistical
sense but also includes the relative belief in that uncertainty
developing over the life of the structure. Examples of this sort
of item include but are not limited to: 1) reliability of the
structure performance with respect to limit state analyses,
structural behaviour, durability, etc., 2) uncertainty of design
loads developing and particularly in regard to load conditions or
combinations, 3) uncertainty of costs associated with design
modifications, geotechnical conditions, and environmental
impacts, and 4) probability of undesirable events.
The design process may be viewed in the feasibility phase
having decisions revolving around financial and environmental
issues (politicians, bankers, government) whereas the detailed
design and construction phases are focused on the technical
engineering and constructability (future Owner/Operator and
Contractor) aspects of the project.
Typically the project design is divided into five phases;
• Feasibility Study – choice whether to construct a tunnel or
not
• Conceptual Design – choice of type of tunnel and alignment
• Preliminary Design – selection of initial design options,
such as overall geometry and material grades
• Detailed Design – selection of final design options
• Construction Design – selection of construction options and
as-builts.
For each level of the design process, the project begins with
an idea and each subsequent phase of the design advances and
requires more detail to be developed. With the completion of
each phase of the project various elements of the design are
established and design options assessed and presented which
allow the “decision-maker” to determine the next course of
action for the subsequent phases of the project. There are
always elements of the design which must be worked out in
detail in the earlier phases of the project, which then will be
required to be properly assessed to evaluate the effects and
resolve issues that will be encountered in later phases of the
project.
5.3.2 Design checks
The design check list given below has attempted to show the
hierarchy of the design steps which are required to complete the
actual design of the structure once the Owner’s requirements
and functionality issues have been resolved.
• List functional requirements
• Select appropriate design codes of practice and design code
protocols
• Determine design loading (empirical or numerical methods)
• Select applied design service loads / load factors / load
combinations
• Perform structural analysis
• Establish required strength / internal forces / moments /
stresses
• Select material strength and reduction factors based on
strength requirements
• Size/re-size structural elements and determine design
strength capacity
• Check design strength capacity against required strength
demand
• Determine serviceability requirements
• Material stress (as applicable)
• Crack control
• Deflections/deformation
• Finalize design based on constructability and material
availability
• Check, check and recheck the calculations and assumptions
Although the above may appear to be common sense, it is
surprising how often errors in the detail of the design have
resulted in large design, construction and contractual problems
for a project. Peer reviews or group review sessions are always
very helpful in identifying shortfalls in any design. Designers
and various team members should learn that these reviews are
geared to identify potential short falls or problems and
determine the most appropriate resolution. The goal is to deliver
the most optimal design for the project.
With today’s computers, often the designers fail to grasp
how the analytical tools currently available for use imply far
greater precision than our ability to understand the ground
properties for which the design must account.
Sensitivity analyses and an ability to understand the potential
hazards and risks associated for the design can help identify
where real savings in the design or real risks in the development
of undesirable performance may exist. Often the sensitivity
analyses will assist in identifying if additional design effort is
really beneficial to the project or not.
In the past, inappropriate decisions in the application of
codes and design methods have led to extraordinary events
which may render a structure useless or far worse create a
potential for injury or loss of life. A recent example is the
 B. Simpson et al. / State of the art Report: Analysis and Design
collapse of the Nicoll Highway excavation in Singapore
(Simpson et al 2008). This issue returns the reader to the
beginning of this section where it is essential that the Owner,
Designer and Contractor clearly understand the functional
requirements of the structure and the performance limits that the
structure demands.
Issues regarding proper modelling of construction sequences
and load development, load combinations, material properties,
serviceability requirements such as deflection/deformation and
crack control as well as selection of appropriate ground and
lining stiffness values are items which can often result in
ambiguity in the definition of applied design loads and may
dramatically affect the resulting performance of the tunnel
structure.
There must always be a clear understanding of how the
design has been undertaken and the construction sequencing
must follow the design or the design must be re-evaluated to
conform to the works in the field. All too often the designer has
predicated the analysis and design on conditions which are
unrelated to the actual works being performed in the field.
Hence, the design is not finished until the tunnel has been
constructed.
5.3.3 The role of geometry
Having the “User” defined requirements identified, there are
always select issues which need to be accounted for in the
design which will adequately allow sufficient space to construct
the final tunnel structure following the excavation and support
phase for SES/NATM type of tunnels.
There are several major design decisions which must be
identified and resolved prior to the excavation phases such as:
• Pressure design requirements (if not environmentally
required)
• Allowance for outer lining installation and deformations
• User requirements such as:
• Ventilation requirements
• Alignment constraints
• Water tightness criteria
• Emergency egress and niches/laybys
• Constructability requirements for excavation and lining
installation
It is extremely important that there is a complete and full
understanding of how these issues relate to the determination of
the tunnel excavation opening size. Once a tunnel excavation is
underway, it is very difficult and costly to have to increase the
size of the existing opening as a result of any design errors or
omissions. This issue has become more important as more
underground works have been following the design-build
project delivery system to accommodate new demanding project
schedules.
5.3.4 The role of maintainability
Regular inspection and maintenance are required to ensure
structures are performing as expected in achieving their
anticipated design life, and to ensure any signs of distress or
deterioration can be diagnosed and addressed in a timely
manner.
The designer must consider how the cut and cover tunnel
will be inspected and maintained over the lifetime of the
structure. For a reinforced concrete box type permanent
structure this is relatively straightforward. However, for cut and
cover construction with integral support of excavation the finish
of the diaphragm wall for highway tunnels in particular may not
be desirable from an Owner’s or user’s perspective.
In such cases one solution has been to disguise the irregular
diaphragm wall finish behind a secondary finish wall.
Obviously the area behind the finish wall is difficult to inspect
and maintain. However, the creation of sufficient space for man
access between the diaphragm wall and finish wall purely for
inspection purposes is expensive. Therefore the secondary wall
is typically placed to leave an air gap of only a few inches
2911
between itself and the diaphragm wall. This gap allows
sufficient space for a camera scope or similar remotely operated
device to be inserted for inspection purposes.
Similar issues surround that of tunnel drain inspection and
maintenance of drained tunnels. This inspection and
maintenance requirement is critical to tunnel performance as the
drains control the magnitude of load the tunnel structure has to
withstand. Drain failure could result in significant over-load of
the structure.
5.4 Design codes and analysis
Unfortunately, there is no known definitive Code identified for
the design of underground openings. Generally designers use
traditional building codes or other codes geared for above-
ground structures and temper the approach and application of
these codes based on underground design experience. Typical
codes that are applied to underground structures are: ACI 318,
ACI 224R-01 (2008), AASHTO (2002), BS8110 (1997),
BS5400 (1988), Eurocodes (EN1992 2004; EN1997 2004) and
other such related design documents.
In 2000, the British Tunnelling Society (BTS) and the
Institution of Civil Engineers (ICE) issued the final draft of their
“Specification for tunnelling” (published by Thomas Telford)
This document refers to the following Eurocodes:
• Eurocode 1 (Basis of design and actions on structures,
1994/95)
• Eurocode 2 (Design of concrete structures, 1992)
• Eurocode 7 (Geotechnical design, 1996).
These codes, subsequently updated as EN1990, EN1992 and
EN1997, have been used in a number of projects in the UK and
throughout the world.
5.4.1 Design code protocol
It is essential to define accurately what codes will be applied,
along with where and how these codes will be used to design
the underground structure. Often misuse or alteration of the
codes and design requirements result in an unfavourable
outcome during the design process. Upon the commencement of
each project, it is extremely important that the protocol for use
of codes is clearly established and defined.
It is the desire of the authors that one day a proper code of
practice will be developed for use in the design of underground
structures and tunnels. Having an established code of this type
will have a distinct advantage of resolving arguments between
Designers, Owners and Contractors as the established code of
practice will provide a sound baseline for the design process.
This aspect becomes ever more important as useable above
ground sites become unavailable, thus promoting the
development and use of underground space.
More than other design works, there are often temporary
structures that are required for the construction of the permanent
structure. In the case of tunnels, this usually exists as braced
excavations or the primary or outer-lining of tunnels constructed
using the SES/SEM or NATM approach. The design protocols
for these structures are no less important than that of the
permanent or inner-lining. The only real exceptions to normal
design requirements is in relation to the potential loading
conditions likely to be encountered over the “useful” life of the
structure and the requisite durability of the structure. Often
requirements for displacement and deformation are not nearly as
stringent as that for the permanent structure provided that the
initial excavation geometry has allowed for these movements so
as not to encroach into the required opening for the final
structure or cause distress to third party utility and land owners.
However, safety to the workers and to third parties must be
maintained at all times and therefore the design of these
temporary works should be subject to similar codes of practice
to ensure that the structure(s) will perform satisfactorily.
2912 B. Simpson et al. / State of the art Report: Analysis and Design
5.4.2 Design standards and codes
The design standard selected for any given application can
either be (a) a material code such as BS8110 or ACI318 which
are principally building related standards, or (b) it can be related
directly to the tunnel function – whether highway or railway
tunnel – in which case in the USA the design would be in
accordance with the American Association of State Highway
and Transportation Officials (AASHTO) or the American
Railway Engineering and Maintenance-of-Way Association
(AREMA) standards respectively.
Table 5.1 indicates the differences in load factors for
controlling loads for several design standards, each of which has
been used in the design of cut and cover structures. By
inspection, the use of one listed standard versus another can
result in significant differences. For example, earth pressure
factors, which constitute one of the largest loads applied to the
structure can vary between 1.4 and 1.7, a difference of
approximately 20%.
Note in particular:
a. AASHTO uniquely allows a factor of 0.65 to be applied for
checking positive moments in frames.
b BS8110 Part 2 allows 1.2 to be used if applied to the “worst
credible” earth and water pressures.
c The values for earth and, especially, water pressure are open
to some interpretation.
Table 5.1. Ultimate limit state load factors
Lateral
Dead Live Hydrostatic
Design Code Earth
Load Load Load
Pressure
AASHTO 1.3 2.17 1.69a
ACI 318 1.4 1.7 1.7
AREMA 1.4 2.3 1.4
BS 5400 1.15 1.5 1.5
BS 8110 1.4 1.6 1.4b
Eurocode 2 1.35 1.5 1.35c 1.35c
While the adoption of any of the above listed standards
results in a functional design, in each case load combinations,
load factors, and material strength reduction factors exhibit
differences which translate into different structure sizes,
different reinforcement quantities and hence different
construction costs.
It is therefore of fundamental importance that the designer
understands the cost implications for the adoption of any one
standard versus another to identify the design standard which
best reflects the function, durability and strength requirements
of any particular project in the most economic fashion.
5.4.3 Design analysis methods
The majority of the analyses undertaken for design of cut and
cover structures and mined tunnels are now computer based.
Analysis models typically comprise two dimensional models
which adequately represent the linear nature of the tunnels.
Three dimensional analyses are typically not necessary but are
useful for modelling specific locations where special conditions
may exist such as rapid changes in cross section, large openings
or appurtenances such as adits, alcoves, and equipment rooms.
A number of software packages are available using either
structurally based software such as Structural Analysis and
Design (STAAD) as manufactured by Research Engineers Inc
which provide for two-dimensional plane frame analysis, or
geotechnically based finite element/finite difference software
such as Plaxis, manufactured by Plaxis BV or Fast Lagrangian
Analysis of Continua (FLAC) as manufactured by Itasca.
Whereas the structural software is obviously focused on the
structural elements of the design, the geotechnical software
places greater emphasis on the simulation of the properties and
behaviours of the surrounding soils and the interaction of the
soils with the structure. Each of the software models and design
approaches has its own benefits and limitations as outlined
below.
For the structurally based software the following findings are
typical:
• Models are simple and quick to develop and amend.
• It is relatively simple to model and simulate large numbers
of individual loads and combinations of loads, using both
service and ultimate load factors.
• Different factors can be applied to earth and hydrostatic
loads based on the likelihood of a given design load to be
exceeded.
• Models must also account for the variations in magnitude
between vertical and horizontal loads as well as accounting
for the loading to be favourable or unfavourable for
structure performance.
• The effects of soil-structure interaction are modelled by
springs which do not represent the ground well. This often
results in the models overestimating the structure forces and
hence the reinforcement requirements
• It is difficult to model adequately the complex sequence of
construction associated with the installation of the support
of excavation system. This typically requires the
introduction of ‘fake’ forces to the models to reproduce the
wall movements from the prior construction stage. This is
particularly important in situations where the support of
excavation is incorporated into the permanent structure.
With the geotechnical software the converse is true. For
tunnel design, given the points identified above, FEM may
1.3 provide a reasonable base to develop and understand the loads
to be used in the design of the tunnel liner. On larger and more
1.4 complex projects with multiple load types, a compromise is to
1.4 use both modelling techniques. The structure software can be
used to develop multiple load cases and combinations to
1.5 identify which cause the most adverse conditions for stress. The
1.4b controlling load combinations from the structural software can
be replicated using the geotechnically based software, thereby
promoting maximum design and construction economy.
5.4.4 Examples of 3D analysis
Yeow and Prust (2005) present an example of use of 3D finite
element analyses to study the construction of a complex tunnel
junction and its effect on existing tunnels. Figure 5.2 shows a
detail of the junction, which was embedded in a mesh of
100,000 finite elements. The construction required about 90
stages of excavation. The ground movement was explicitly
computed from the 3D tunnelling model without having to
resort to the introduction of volume loss through stress
relaxation or imposed volume change. Computed deformation
were used for damage assessment. Ability to prepare data for
analysis of this type and to carry it out in practicable timescales
is still developing very rapidly. However, as discussed by
Yeow and Prust, providing adequate models of the soil and
structural materials remains a major challenge.
Further examples are presented by Simpson et al (2006).
Figure 5.3 shows a situation at Stratford in East London, where
the new Channel Tunnel rain Link tunnel was to pass beneath
exiting metro tunnels of the Central Line. Since it is known that
prediction of settlement due to tunnelling is extremely difficult,
a special method was devised in which an empirically derived
settlement field was imposed on a 3D finite element mesh
which included the exiting tunnels. The details of the method
are presented by Yeow et al (2005). Figure 5.4 shows the
computed deformation of the existing tunnel which had a bolted
steel lining, from which bolts were to be removed at various
points. It was predicted that if the design ground loss occurred
in the new tunnels the existing tunnel would not bend much but
would shear at sections where the bolts were removed. In the
event, actual ground loss and distortion of the existing tunnels
were relatively small.
 B. Simpson et al. / State of the art Report: Analysis and Design 2913

5.4.5 Example – sprayed concrete lining in clayey silt take up load. A particularly important decision is the choice of
The example shown in Figure 5.5 illustrates some of the overburden pressure, that is, whether any allowance is made for
problems to be considered in analysis of a sprayed concrete the type of arching action suggested by the finite element
lining. The internal tunnel dimension is approximately 11m analysis. Duddeck and Erdmann (1982) recommended that for
wide by 10m high, with a cover to diameter ratio (C/D) less shallow tunnels, C/D<3, a model without reduction of ground
than 2.0 It was excavated using an upper heading with pressure at crown is appropriate. In general, for shallow tunnels
temporary invert, with the bench excavated some distance (C/D < 2) the full overburden assumption is usually made.
behind the top heading. The primary lining of sprayed concrete Such an approach leads makes much greater demands on the
is 350mm thick, and the secondary lining of cast-in-situ capacity of the tunnel lining than does reliance on the results of
concrete is also 350mm thick. a finite element analysis of the type shown in Figure 5.8.
Figures 5.6 and 5.7 show two forms of analysis which have
been considered: finite elements and beam-spring models. The
2D finite element analysis uses a simple elastic-Mohr-Coulomb
model and takes the mesh through all the stages of tunnel
construction – heading and invert – with an appropriate
allowance for ground loss assumed to occur during the
excavation process. This inevitably leads to reduction in the
final stresses on the tunnel lining as the soil arches over the de-
stressed area of the tunnel, as illustrated in Figure 5.8. It is
assumed that in the long term the primary lining degrades and is
unable to carry bending stresses. If required, the formation of
plastic hinges in the linings, both primary and secondary, can be
modelled, though careful inspection of any plasticity which
implies cracking in the secondary lining is needed by designers
concerned about durability. Particular attention should be paid
to the rotation capacity at the cracked joint.

Figure 5.3: Intersection of the Channel Tunnel Rail Link and Central Line
tunnels at Stratford, East London.

Figure 5.4: Computed deformation of a Central Line tunnel.

Figure 5.5: Tunnel construction in clayey silt

Figure 5.2: 3D modelling of SCL tunnelling operation (after Yeow and

Prust 2005)

The spring model shown in Figure 5.7 can be used for an

analysis following the general procedures recommended by
Schulze and Duddeck (1964). This requires the designer to take
decisions about the stress field within which the tunnel is
placed, both in terms of the vertical and horizontal stresses,
though these are modified as the lining flexes and the springs Figure 5.6: Detail of 2D mesh at the tunnel
2914 B. Simpson et al. / State of the art Report: Analysis and Design
Figure 5.7: Beam and spring model
Figure 5.8: Principal stress plot showing arching
A further issue for both types of analysis is to decide how
factors of safety, particularly load factors, should be applied.
As noted in Table 5.1, the codes require that different factors
are applied to dead and live loads. Some of them also
differentiate between “unfavourable” and “favourable” dead
loads, applying a factor of 1.0 to favourable dead loads. These
distinctions are easily made if the factors are applied to the
input data of the computations, but more difficult if an
unfactored analysis is carried out, with the intention of factoring
the resulting bending moments and thrusts.
Applying global factors to the forces derived from an
unfactored analysis may significantly underestimate the adverse
effects of live loading upon the structure. For instance the
transient effects associated with heavy rain upon partially
saturated soils may be underestimated if factors are simply
applied to the loads derived from a calculation based on the soil
properties in the partially saturated state. In that instance the
adverse combination of both a reduction in strength and
stiffness of the soil and a temporary increase in load may not be
fully accounted for.
Fortunately, live loading is often less important to the design
of tunnel linings, but the distinction between favourable and
unfavourable dead loads can be controversial. Much of the
loading on a tunnel lining is due to the difference between
vertical and horizontal stresses in the ground. If one of these is
deemed to be favourable whilst the other is unfavourable, the
computed bending moments in the lining are increased
considerably. More common practice is to regard the ground
loading as coming from a “single source” (in the terms of
EN1990), so that the same load factor is applied to all earth
pressures. If this is the case, and if live loading is unimportant,
the effect of applying load factors to input data of the analysis,
or to the resulting bending moments, may be the same.
5.5 Materials and sustainability
In fairly recent years the use of cement replacement materials
such as ground pulverised fuel ash (PFA) and granulated blast
furnace slag (GGBFS) has become commonplace in the
specification of reinforced concrete for cut and cover and other
forms of tunnel. These materials are by-products from industrial
practices from the power and steelmaking industries
respectively.
The use of these replacement materials in reinforced
concrete has been demonstrated to offer a number of advantages
over Portland cement in terms of the durability and long term
performance of the concrete. These advantages include:
• a denser concrete mix which improves watertightness
• reduced requirement for mix water
• increased resistance to chemical attack
• reduced heat of hydration during setting and curing
The final bullet point is particularly significant. Lowering
the heat of hydration minimizes the temperature difference
between ambient air temperature and the peak temperature
within the concrete matrix during the setting process. By
minimizing the temperature differential the incidence of
problematic thermal cracking can also be minimized. This is of
particular importance for massive concrete structures such as
cut and cover structures which feature large concrete pours and
onerous conditions of restraint at wall/slab interfaces as thermal
cracks in theory will extend through the entire thickness of the
concrete section.
However, there are some disadvantages to the use of cement
replacement materials. The rate of gain of strength is slower
than for Portland cement concrete, which results in forms being
left in place for longer; material properties can vary
necessitating that all supplies come from a uniform source
where the properties are well understood. Analysis by Yazdchi
et al (2005) suggested that the speed of construction of a small
diameter tunnel with a sprayed concrete lining could be
restricted by lower rate of gain of strength.
It can be seen from Table 5.2 that while there is extensive
use of PFA in Hong Kong, Japan and the UK, in other major
construction markets considerable room for growth in the use of
cement replacement materials exists.
Table 5.2: PFA Production and Utilization 1995, after Meyer (2005)
except UK data which is more recent, after Barnes and Sear (2004).
Country Million Tons Million Tons % Utilization
Produced Utilized
China 91.1 13.8 15.1
Denmark 1.3 0.4 30.8
Hong Kong 0.63 0.59 93.7
India 57.0 2.0 3.5
Japan 4.7 2.8 59.6
Russia 62.0 4.3 6.9
USA 60.0 8.1 13.5
UK 5.7 3.1 55
In terms of the long term performance of underground
structures the use of cement replacement materials is
recommended. In addition, as natural resources become scarcer
and more expensive, the expanded use of cement replacement
materials should be promoted.
The use of fibres, such as steel or polypropylene, in concrete
for crack control and handling or to prevent explosive spalling
of concrete under fire or an explosive event, respectively, is
increasing throughout the world. The past performance in early
projects has supported the industry-wide use of these fibres and
as such, use of fibres is now given significant recognition and is
widely employed on modern day projects.
5.5.1 Instrumentation and communications
While there has been widespread use of geotechnical
instrumentation to monitor ground movements before and
during construction, with the advent of the internet has “real
time monitoring” – web based data provision and retrieval –
many new systems are being developed everyday.
 B. Simpson et al. / State of the art Report: Analysis and Design 2915

These systems offer a distinct advantage to construction of face of about 40m to 150m, a bench is taken out below the top
underground works in that immediate notification of adverse heading and supported. The final excavation of the floor, or
performance may be transmitted to the design/construction team invert, is taken out and supported, fairly close to the bench face,
24 hours a day. This level of monitoring and notification should depending on the ground conditions. The closed support ring
allow corrective actions to be implemented in a timely manner, around the full tunnel section comprises the outer lining, which
thus averting undesirable events. must stabilize the excavation before the inner lining is placed.
The outer linings generally have not been considered as part of
5.5.2 Performance monitoring the long-term or permanent support of the tunnels.
Tunnel excavation and lining installation should always include During excavation, survey displacement monitoring stations
both outer lining and inner lining monitoring programs. For cut- should be installed close to the excavation face, typically at
and-cover structures this would correlate to support of intervals of 10m to 20m, to measure the settlement and
excavation systems and the permanent structure, whereas convergence of the outer lining and provide verification of the
monitoring devices are located for not only the temporary works design of the outer lining. In poor ground conditions the spacing
but for the permanent works as well. By default, ground loading may be reduced to as little as 5m. Generally, each station
on the tunnel is directly related to the deformation of the ground consists of five targets, three in the top heading and two in the
and accurate and precise monitoring is one of the few tools bench. The monitoring results should be used to verify the
available to verify the design assumptions regarding load and design assumptions for the outer and inner linings. An
stability of the excavation. important aspect is that the information will be used to verify
The monitoring program can also include instrumentation to the assumed design ground loadings for the inner linings. Refer
better understand ground movement beyond the limit of the to Figure 5.9 for typical instrumentation layout.
excavation as well as loading on the outer and inner liner.
Distinct instrumentation stations that are representative of
sections of expected ground types (classes) are extremely
helpful in verification of anticipated tunnel performance as well
as for areas that have been determined to be difficult or
problematic.
It is advantageous that the Designer specifically requires that
these instrumentation stations are installed and that locations of
installation are confirmed and agreed to by the Designer.
Regular reading, interpretation and evaluation of the stations is
also paramount. In mined tunnels, often the full ground load
resulting from the excavation is not evident until the drive face
is 4-6 diameters past any given location.
Predicted deformations of the outer lining as provided by
analysis may be selected and used as “control” or “target”
values during the excavation whereby these are compared to the
measured deformations from the monitoring stations.
Depending on the ratio of the calculated deformations to the
measured ones, certain pre-agreed actions are implemented
when the ratio reaches certain values. For example, “alert level
1” occurs if the ratio reaches 1/3, “alert level 2” if 2/3, and
“alert level 3” if 1.0. The control values are determined for each
stage of the excavation thus allowing for deformations to occur
for each of the 3 standard stages of excavation, heading, bench
and invert. A typical relationship between the “Control Criteria” Figure 5.9 – Instrumentation Station
and “Safety Control System” is provided in Table 5.3.
5.5.3 Constructability reviews
Constructability reviews are as essential to the design as are the
Table 5.3 Lining Performance Control Criteria drawings, specifications and supporting calculations. All the
drawings and calculations will be for nothing if the design is not
CONTROL
ALERT LEVEL
VALUE
ACTION constructible. As an example, if difficult ground is encountered
the tunnel invert excavations should be made deeper and more
Standard Operations - Standard Measurements circular; this should be anticipated in cases where ground
conditions are not well known or if it is known that poor ground
Increase Measurement conditions are likely to exist. This sort of flexibility in the
1/3 Control Frequency, Conduct Site design should be considered in the beginning and properly
Level I Alert
Value Inspections, Issue Strict planned for, including layout of plant and equipment. The
Work Instructions general concept of excavation equipment should be flexible to
Strengthen Measurement meet the changing demands of underground work during the
2/3 Control System, Carry Out / Perform tunnel construction.
Level II Alert
Value Minor Control Works Optimized reinforcement, especially in the arch lining,
(“Auxiliary Measures”) benefits the inner lining. High reinforcement content will
Stop Excavation, Analyze increase the probability of honeycombing and voids. These
Control Cause and Tendency of reviews are often performed prior to tender award but they also
Level III Alert
Value Deformation, Select Tunnel continue through the construction phases as well.
Reinforcement Measures The beginning of mined tunnelling should always be based
on a conservative design. The observations during early site
In view of the large excavation sizes, the tunnel cross-section execution should accelerate the learning curve and optimize
is usually excavated in stages. Initially a top heading with an support requirements. As information is gathered, the Tunnel
invert on or just above the tunnel spring line is excavated and Construction Engineer and the Lead Geotechnical Engineer can
stabilized, if necessary with a temporary invert of shotcrete to
form a closed support ring. At a variable distance behind the
2916 B. Simpson et al. / State of the art Report: Analysis and Design
work closely with the Designer who can then refine the analysis
to support the decisions for lining design and support selection.
Precise Monitoring of deformation and interpretation of the
deformation will help identify the best support solutions to
minimize risk of collapse. However, deformations alone do not
always tell the full story. Full instrumentation stations which are
capable of identifying ground mobilization and lining stress
allow for a full understanding of the interaction between ground
strains and lining performance.
5.6 Future developments
In this emerging era of climate change and constrained
resources it is incumbent on the engineering profession to
design for value through economic use of materials and the
promotion of engineered solutions which are durable and
sustainable.
5.6.1 Design standards and codes
Development of a standard specific to below grade construction
which is consistent with construction practices (or a number of
standards dependent upon regions/jurisdictions) will help the
review and acceptance of designs and construction performance.
As an example, appropriate load factors for groundwater
levels are an important element of the design when hydrostatic
pressures form a large part of the permanent load. The larger the
percentage of total load resulting form water pressure should
merit high scrutiny of the selection of load factor to be used
with that load. Long term creep and concrete strain need to be
reviewed in relation to the load applied..
Accordingly the UK National Annex to Eurocode 7
allows/encourages “direct assessment” of design values of
ground water pressure (implying that no further factor is to be
applied to these):
The partial factors specified in the National Annex to BS EN
1990:2002 [ie the normal partial factors on loads] might not be
appropriate for self-weight of water, ground-water pressure and
other actions dependent on the level of water, see 2.4.7.3.2(2).
The design value of such actions may be directly assessed in
accordance with 2.4.6.1(2)P and 2.4.6.1(6)P of BS EN 1997-
1:2004. Alternatively, a safety margin may be applied to the
characteristic water level, see 2.4.6.1(8) of BS EN 1997-1:2004.
An alternative approach is to calculate structural stresses for
unfactored “characteristic” water pressures and apply factors to
the structural stresses. There is considerable merit in checking
both approaches in a design.
Accordingly design standardization and consistency of
application of load factors/combinations etc. are very important
aspects of design for underground structures and further
development of the codes for design of such structures is
warranted.
5.6.2 Analysis
More widespread integration of the temporary support of
excavation (SOE) into the permanent structure helps to reduce
environmental impacts associated with the work as it lessens the
tunnel footprint and minimizes impacts upon utilities, and
expensive real estate acquisition.
Soil-structure interaction models can be used to demonstrate
the ability of SOE in the long term condition to relieve load on
the cast in place structure, and correspondingly reduce structure
size and cost. Use of integral SOE should be studied further to
assist in the use of a hybrid solution for the permanent structure,
while avoiding potential durability issues.
5.6.3 Materials
It is well documented that the concrete industry is one of the
largest consumers of natural resources – relatively recent
statistics suggest that the concrete industry consumes over 10
billion tons of sand and aggregates and 1 billion tons (1 trillion
gallons) of water, not including water for wash down of mixers
or curing on an annual basis (Transportation Research Board
2007, Mehta 2004, Meyers 2005).
In addition to the depletion of these natural resources, the
production of cement in itself expends considerable amounts of
fossil fuel and electrical energy. Annual production of cement is
approximately 1.6 billion tons, and is expected to exceed 2
billion tons by 2010. The energy expended on the creation of
one ton of cement generates an equivalent weight of carbon
dioxide (CO2), a principal contributor to global warming. The
annual production of 1.6 billion tons (and rising) of CO2
corresponds to approximately 7% of the global emission of this
gas.
In addition, concrete construction debris from demolition
constitutes a large percentage of solid waste disposal. In North
America, Europe and Japan concrete and masonry rubble
accounts for approximately two thirds of all construction and
demolition waste. It is reported that in excess of 1 billion tons of
construction and demolition waste is generated each year
(Mehta 2001).
Clearly none of these practices is sustainable in the long
term. However, solutions to mitigate or at least reduce their
impacts are being identified and implemented. These solutions
include the increased use of cement replacement materials – as
identified earlier, these materials are significantly underutilized,
and the use of recycled concrete aggregates should be promoted.
Typically concrete mixes are specified to contain roughly
30% of cement replacement material. With the benefits
described previously proponents of the use of pfa/ggbfs in
concrete production advocate that this percentage should be
significantly higher and a content of 50-60% cement
replacement should be sought. This increased volume of cement
replacement materials will reduce requirements for cement
production and correspondingly reduce CO2 emissions.
An additional benefit of increased usage of cement
replacement materials is the reduced requirement for mixing
water. Concrete mixes using high percentages of cement
replacement material have been demonstrated by testing to
require 20% less mix water than corresponding mixes which are
purely cement based. In many regions of the world water is a
precious resource. A 20% reduction in concrete mix-water
would equate to approximately 200 billion gallons of water
saved per year. This is undoubtedly a significant volume (Mehta
2004).
5.6.4 Contracting practices
Due to the complexities involved with underground
construction in an urban environment, it is time to move away
from low bid construction which is prevalent over much of the
world to a ‘best value’ approach whereby the contractors
technical proposal forms an official part of the bid evaluation
process (Pollalis 2006).
Consideration should be given to early contractor
involvement in the design process, using a negotiated or target
price contract approach as was successfully implemented on the
UK Channel Tunnel Rail Link (Cathart 2005).
Under such a scenario qualified contractors would bid on
design documents complete to a preliminary engineering level
of detail. The selected contractor would subsequently work with
the owner and designer to complete the design engineering.
Through this process, the designer understands and can design
to accommodate the contractor’s proposed means and methods,
the contractor provides constructability review of the design,
and the owner receives a bid price with reduced contingency.
Once the design is finalized, the owner and contractor
renegotiate the contract price based upon the improved
understanding and development of the project scope. In the
event that an agreement cannot be reached, the parties can
terminate negotiations, the contractor is paid for his design
effort, and the contract can be rebid.
 B. Simpson et al. / State of the art Report: Analysis and Design
6 SEISMIC THEMES OF GEOTECHNICAL
ENGINEERING INTEREST: BEYOND THE STATE OF
PRACTICE
6.1 Introduction: topics that have emerged from recent
earthquakes
This part of the report addresses issues related to what is called
“performance-based design” of foundations against two
earthquake−related hazards: (i) emergence of the rupture of a
seismic fault underneath a structure, and (ii) strong dynamic
shaking resulting from seismic waves emanating from the whole
rupturing fault. Against the latter hazard (i.e. strong ground
shaking) emphasis will be on the mobilisation of bearing
capacity mechanisms for slender structures on shallow
foundations. The topics chosen in this chapter have been
prompted by observations in many recent earthquakes, most
importantly from Northridge (1994), Kobe (1995), Kocaeli
(1999), and Chi-Chi (1999). Valuable lessons of great
significance have emerged from these events. Here are a few of
them:
• Structures on conventionally stiff raft and box foundations
can survive dip-slip and strike-slip fault ruptures producing
offsets of the order of 1m beneath the structure. By
contrast, pile foundations are directly affected by the rupture
and, moreover, tend to impose on the super-structure the
fault-induced differential displacements; both structural and
pile damage are thus likely to occur.
• Even on very soft soil, a slender structure subjected to
strong shaking may undergo severe rocking oscillations,
accompanied by uplifting. Large deformations are thus
induced in the soil and bearing capacity mechanisms are
mobilised alternately under each half of the foundation. But
the structure does not necessarily topple, and perhaps it does
not even suffer detrimentally-large residual rotation and
displacement.
• Geotechnical systems yielding only in one direction, such as
gravity retaining walls and slopes, are quite sensitive to the
nature and direction of seismic shaking and are thus greatly
affected by phenomena such as forward-directivity and
fling-step, often observed close to the seismogenic fault.
• Conventionally designed caisson quay-walls prove to be
rather robust structures against toppling but they may
experience (large) seaward displacement and rotation if the
foundation soil is very deformable or, even worse,
liquefiable; by contrast, the retained soil does not liquefy
(contrary to a widely-held misconception) and will only
exert moderate forces on the wall, of the same order as the
pseudo-statically determined Mononobe-Okabe active
forces.
• End-bearing pile foundations can perform very well in
“level” ground, even if the surrounding soil were to liquefy
massively, as long as there is no danger of buckling (usual
case with the present-day large diameters of cast-in-place
piles). However, in “non-level” ground, liquefaction
triggers soil flow; as a consequence, piles could undergo
severe “kinematic” lateral deformations and possibly
bending failure. (“Kinematic” refers to ground-
displacement induced stressing of the pile along its depth, as
opposed to “inertial” loading which originates at the
superstructure and is then transmitted on the top of the
piles.) The possibility of buckling of small diameter end-
bearing piles cannot be dismissed, however, as has been
pointed out by Bhattacharya et al (2004, 2005) and Kerciku
et al (2007).
Many theoretical and experimental studies have been
initiated following these field observations, so that today the
range of topics of interest in Seismic Geotechnics has greatly
2917
expanded. Two such topics are highlighted in this paper. The
emphasis is on elucidating the analysis of seismic soil-
foundation interaction in the presence of large soil deformations
and near-failure conditions.
It is emphasized, however, that many of the ideas outlined in
the following sections, although supported by some field
evidence, are not yet to a sufficient extent developed to be
directly applicable in design. Many more analytical studies and
several large-scale or centrifuge experimental tests, as well as
additional well-documented field evidence, are needed before
these ideas are turned into reliable design methods accepted by
the profession.
6.2 Foundations interacting with a rupturing seismic fault
6.2.1 Problem and motivation
“Strong” structures founded on the surface of, or at depth in,
soil have often resisted successfully the loading induced by a
rupturing seismic fault (Duncan & Lefebvre, 1973; Bray 1990).
In the three Turkey and Taiwan earthquakes of 1999 numerous
structures (buildings, bridge piers, retaining structures,
electricity pylons, dams, tunnels) were located directly above
the propagation path of the rupturing faults ⎯ strike slip,
normal, reverse. Some of these structures performed
remarkably well; others failed dramatically. These observations
had a strong motivating influence to modify the pertinent
clauses of seismic codes and to conduct further research. For it
became immediately clear that the strict prohibition of building
in the immediate vicinity of active faults, which the prevailing
seismic codes have demanded, was unduly restrictive if not
meaningless.
In addition to several geologic factors that contribute to such
behaviour, the role of a soil deposit that happens to cover the
rock appears to be significant. If, where, and how large will the
dislocation emerge on the ground surface (i.e. the fault will
outcrop) depends not only on the style and magnitude of the
fault rupture, but also on the geometric and material
characteristics of the overlying soil deposits. Field observations
and analytical and experimental research findings (Bray et al,
1994a, b; Cole & Lade, 1984; Lade et al, 1984; Anastasopoulos
et al 2007) show that deep and loose soil deposits may even
mask a small-size fault rupture which occurs as a dislocation at
their base; whereas, on the contrary, with a cohesive deposit
especially of small thickness, a large offset in the base rock is
likely to cause a distinct fault scarp of nearly the same
displacement magnitude. One important finding of the above
studies is that the rupture path in the soil is not a simple
extension of the plane of the fault in the base rock: phenomena
such as “diffraction” and “bifurcation” affect the direction of
the rupture path, and make its outcropping location, the offset
magnitude, and the shape of the deformed surface difficult to
predict.
Figure 6.1: Configuration of the soil–foundation system subjected to a
normal fault dislocation at the base rock.
Our main interest here is to study how a structure sitting on
top of the fault breakout behaves. It turns out that an interplay
occurs between the propagating fault rupture, the deforming
foundation soil, the differentially displacing foundation, and the
2918 B. Simpson et al. / State of the art Report: Analysis and Design
supported structure. Two different phenomena take place.
First, the presence of the structure modifies the rupture path.
Depending on the rigidity of the foundation and the weight of
the structure, even complete diversion of the fault path before it
outcrops may take place. Obviously, the damage to a given
structure depends not only on its location with respect to the
fault outcrop in the “free-field”, but also on whether and by how
much such a diversion may occur. Second, the loads
transmitted from the foundation on to the soil tend to compress
the “asperities” and smoothen the “anomalies” of the ground
surface that are produced around the fault breakout in the
free−field, i.e. when the structure is not present. Thus,
depending on the relative rigidity (bending and axial) of the
foundation with respect to the soil, as well as on how large the
structural load is, the foundation and the structure experience
differential displacements and rotation, different from those of
the free-field ground surface. This phenomenon, given the
name Fault Rupture−Soil−Foundation−Structure Interaction
(FR−SFSI) is briefly elucidated below.
6.2.2 Shallow foundations on top of rupturing fault
The problem studied here is illustrated in Figure 6.1. A uniform
soil deposit of thickness H at the base of which a normal fault,
dipping at an angle Į (measured from the horizontal), produces
downward displacement (called “dislocation” or “offset”) of
vertical amplitude h. Note that the movement of the fault
during an earthquake is in one direction only (and not
oscillatory) and takes place rather slowly (on the order of tens
of seconds rather than one-tenth of a second) − Ambraseys and
Jackson (1984). The analysis is static and is conducted in two
steps. First, fault rupture propagation through soil is analysed
in the free field, ignoring the presence of the structure. Then, a
strip foundation of width B carrying a particular superstructure
is placed on top of the free−field fault outcrop at a specified
distance S (measured from its corner), and the analysis of
deformation of the soil–structure system due to the same base
dislocation h is performed. The analyses are conducted under
2-D plane-strain conditions ⎯ evidently a simplification, in
view of the finite dimensions of a real structure in the direction
parallel to the fault. A limited number of 3-D analyses have
shown that the 2-D approximation leads to slightly conservative
results (i.e., it leads to somewhat larger rotation) (Gazetas et al
2007).
Among several alternatives that were explored, the FE model
shown in Figure 6.2 produced results in excellent accord with
several centrifugal experiments conducted at the University of
Dundee for both steps of the analysis (Anastasopoulos et al
2007, 2009). A parametric investigation revealed the need for a
long (B ≈ 4H) and very refined mesh (element size of 0.5 m −
1.0 m) along with a suitable slip-line tracing algorithm in the
region of soil rupture and foundation loading. An elastoplastic
constitutive model with the Mohr-Coulomb failure criterion and
isotropic strain softening was adopted and encoded in the
ABAQUS finite element environment. Similar models have
been successfully employed in modelling the failure of
embankments and cut slopes (Potts et al, 1990). Modelling
strain softening was shown to be necessary; it was introduced
by suitably reducing the mobilised friction angle ijmob and the
mobilised dilation angle ȥmob with increasing plastic octahedral
shear strain. With all the above features, the FE formulation is
capable of predicting realistically the effect of large
deformations with the creation and propagation of shear bands.
The foundation, modelled with linear elastic beam elements,
is positioned on top of the soil model and connected to it
through special contact elements. The latter are rigid in
compression but tensionless, allowing detachment of the
foundation from the bearing soil (i.e. gap formation beneath the
foundation). The interface shear properties follow Coulomb’s
friction law, allowing for slippage. Both detachment and
slippage are important phenomena for realistic foundation
modelling.
Figure 6.2: Finite element discretisation and the two steps of the
analysis: (a) fault rupture propagation in the free-field, and (b) interplay
between the outcropping fault rupture and the structure (termed Fault
Rupture–Soil–Foundation–Structure Interaction, FR-SFSI).
A typical result showing the interplay between loose (ID =
45%) soil, rupture path, and a perfectly rigid foundation
carrying a 3-storey structure is given in Figure 6.2. A base rock
dislocation of 2 m (5% of the soil thickness) is statically
imposed. The structure is placed symmetrically, straddling the
free-field fault breakout (i.e. the foundation is placed with its
middle coinciding with the location where the fault outcrops in
the free field). A distinct rupture path (with high concentration
of plastic shearing deformation and a resulting conspicuous
surface scarp) is observed only in the freefield. The presence of
the structure with its rigid foundation causes the rupture path to
bifurcate at about the middle of the soil layer. The resulting two
branches outcrop outside the left and near the right corner of the
foundation, respectively. The soil deformations around these
branches are diffuse, and the respective surface scarps are much
milder than in the free field. Thanks to the substantial
weight of the structure and the flexibility of the ground, the
structure settles and rotates as a rigid body. The foundation
does not experience any loss of contact with the ground;
apparently, the foundation pressure is large enough to eliminate
any likely asperities of the ground surface. As a result of such
behaviour, the structure and its foundation do not experience
any substantial distress, while their rotation and settlement
could be acceptable.
Several tectonic, geometric, and material factors affect the
interplay between an emerging fault rupture, the soil, and the
foundation. For a fairly detailed parameter investigation
relating to the subject, see Anastasopoulos et al (2007, 2009).
This presentation will consider the consequences of this
interplay for a rigid mat foundation, 20m wide, carrying a 2-
story frame structure, and resting on top of a 40m thick deposit
of dry sand. Two different densities are considered
parametrically: ID = 45% (loose sand) and ID = 80% (dense
sand). Typical values of (peak and critical-state) angles and
dilatancy angle are assigned to each density, in addition to some
other secondary modelling differences. For ID = 80%: ijp =
45Ƞ , ijcs = 30Ƞ , ȥp = 18Ƞ, ȥcs = 0o. For ID = 45%: ijp = 32Ƞ ,
ijcs = 30Ƞ , ȥp = 3Ƞ, ȥcs = 0o. Regarding their stiffness, both
soil deposits are taken as "Gibson” type soils, i.e. with elastic
Young’s modulus increasing linearly with depth: E = 5z (MPa,
m) for ID = 80% and E = 2z (MPa, m) for ID = 45%. The style
and magnitude of the fault rupture play an important role in the
response (and eventually the survival) of the structure. A
normal faulting is considered here, emerging with an angle Į =
45o on the surface of the base rock and with a dislocation
(offset) having a vertical component h = 2m, that is 5% of the
overlying soil thickness (H =40m).
Another important factor is the exact position of the
foundation with respect to the outcropping location of the fault
rupture on the ground surface. Since this location is affected by
the foundation-structure itself, we use as reference the point O
 B. Simpson et al. / State of the art Report: Analysis and Design
in Figure 6.1, where the fault outcrops at the free field ground
surface. Having numerically determined point O, we specify as
the position of the foundation the distance S of its left edge (i.e.
the edge on the hanging wall) from O.
Two values of S are considered here: S = 4m and S = 16m.
In the former case, the fault, if unaffected by the foundation,
would have emerged near the left edge of the foundation; 80%
of the foundation would have been located on the stable
footwall. In the latter case, the fault would have emerged near
the right edge of the foundation; only 20% of the foundation
would have been located on the stable footwall. Figures 6.3 and
6.4 illustrate the response of the foundation in these two cases,
for each of the two soil densities. (In both figures the vertical
scale is substantially exaggerated; the slopes thus appear much
steeper than in reality.) It is noted that in all four cases
examined the foundation is lightly loaded (mean contact
pressure po = 30kPa). The influence of po on the behaviour of
the foundation-structure system will be explored later.
Several trends are worthy of note in the two figures:
1. For the first case, of the fault emerging near the left edge (S
= 4m, or S/B = 0.20), this lightly loaded foundation causes
only a minor diversion of the rupture path from its free-field
position (Figure 6.3). This diversion is clearly noticeable
only in loose soil (deviation to the left of about 1.5 m away
from the footing centre, i.e. towards the hanging wall).
2. A profound consequence of the emergence of the fault
under the foundation is the development of gaps between
soil and foundation. On the dense soil nearly the whole left
part of the foundation ( ≈ 4 m long) turns into a cantilever.
The structural loads above this part induce a (substantial)
Figure 6.3: Response of soil surface and foundation due to a normal fault emerging at a distance S = 0.20B from the left
edge under the foundation. (Vertical scale exaggerated.)
Figure 6.4: Response of soil surface and foundation due to a normal fault emerging at a distance S = 0.80B from the left
edge under the foundation. (Vertical scale exaggerated.)
2919
bending moment ⎯ this indeed constitutes the major
distress of the foundation from the rupturing fault. The
survival of the structure depends on its ability to safely
sustain this moment.
3. In dense soil, the observed asymmetric reduction in the
contact area leads to an increase of the average normal
contact pressure (to about 1.25 po) and to an unavoidable, if
small, rotation of the foundation. All this culminates in a
nearly triangular distribution of normal contact pressures
with a peak of about 3po near the edge of the fault scarp,
while the right edge of the foundation starts to uplift from
the soil. This generates an additional cantilever at the right
edge and further aggravates the rigid-body rotation of the
foundation. This constitutes the second, “operational”
distress of the foundation from the rupturing fault. Note,
however, that for large enough values of po the local soil
yielding will reduce the peak and lead to a more uniform
pressure distribution.
4. On loose soil the size of the gap is restricted to 2.5 m, while
along the remaining 17.5 m of the foundation width full
contact is maintained. Hence, the bending moment of the
cantilevered part is only 40% of the bending moment that
develops under the same conditions on the dense soil. This
favourable behaviour stems apparently from the greater
compressibility of the loose soil and the ensuing depressing
of the fault scarp, as is evident in the figure. However, an
unfavourable consequence of the increased soil
compressibility is the greater (by a factor of nearly 3)
rotation of the rigid foundation.
2920 B. Simpson et al. / State of the art Report: Analysis and Design

2. For the second case (Figure 6.4), of the fault emerging
near the right edge (S = 16m or S/B =0.80), a substantial
diversion of the fault takes place in both loose and dense
sand: the fault deviates by 4 m towards the footwall in
both cases, emerging beyond the foundation at the right-
hand side. This significant theoretical finding has been
verified with several observations of actual behaviour in
the Kocaeli 1999 earthquake (Anastasopoulos & Gazetas
2007a,b). But the similarity in behaviour between loose
and dense sand ends here. Significant differences are
noted between the two cases:
• The fault scarp that is formed near the right edge of the
building is far more conspicuous in loose sand.
• In dense sand the fault rupture undergoes bifurcation,
with the secondary rupture branching to the left and
emerging underneath the structure, not far from its centre.
• As a result, on dense sand, the middle part of the building
loses contact with the bearing soil for about 10m, while
the left and right part remain in contact for about 2m and
7m, respectively.
On loose sand, the response of the foundation is quite
favourable: not only is the dislocation diverted and outcrops
beyond the right edge as already noted, but full contact is
maintained over the whole length of the soil−foundation
interface. The distress of the foundation is thus significantly
less on loose than on dense sand. Also smaller on loose sand
is the (rigid-body) rotation of the foundation, thanks to the
larger depressing of the fault scrap on the more compressible
soil. Such a good response of a building founded on loose soil
on the hanging wall is reminiscent of several success stories
from the Kocaeli 1999 earthquake, especially of the building
in Denizevler across the entrance from the Ford factory, near
Gölcük (see Anastasopoulos et Gazetas 2007a). Figure 6.5
shows in some detail this building and the geometric
characteristics of fault outcropping.
The above results are all for a specific small contact
pressure, po = 30 kPa, at the foundation−soil interface, typical
of a 2 story building. Although not shown here, the effect of
an increase in the number of stories is quite beneficial on
loose sand, and somewhat less on dense sand. The most
significant benefits are the decrease of foundation rotation
(and thereby of building tilting) and the elimination of a large
part of uplifting. As a consequence, the survival of a heavy
building founded on loose soil above a major fault rupture
seems possible, in qualitative accord with numerous such
success stories in several earthquakes.
The significance of the magnitude of the (average) contact
pressure, po, is summarized in Figure 6.6. The figure gives
parametric results for the size of the unsupported (detached)
regions, u, and the effective (in-full-contact) regions, b, of the
foundation. Both are normalized by the width B of the
foundation. In addition to po, the examined variables include
the location of the foundation with respect to the free-field
fault outcrops, S, and the relative density of the soil, ID.

Figure 6.5: The remarkable performance of a 5-story building near Gölcük, Turkey, “on top” of the normal fault rupture in the Kocaeli 1999
Earthquake: (a) photograph showing the fault being directed towards the building; (b) photograph showing the vertical displacement
reaching 2.3 m, along with a horizontal component of 1.1 m, measured on the torn-apart fence of the building; (c) simplified cross-section of
the building and the fault; and (d) plan view of the foundation (box-type foundation with cross tie beams), along with the horizontal
displacements measured around the building, and a sketch of the diverted fault. (after Anastasopoulos & Gazetas, 2007a).
 B. Simpson et al. / State of the art Report: Analysis and Design 2921

Figure 6.6: Summary of parametric results showing in normalised form which parts of the foundation remain in full contact
(constituting the effective width, b, or b1 + b2 in certain cases) or separate from the soil (constituting the unsupported width, u,
or u1 + u2 in certain cases). The fault ruptures at three different locations: (a) s = 0.2B, (b) s = 0.5B, and (c) s = 0.8B. The effect
of surcharge load, po, is shown for the dense (left) and the loose (right) soil deposit. Indicative cross-sections of the foundation
on dense and loose sand are shown on the left and on the right, respectively, showing the no-contact areas and crude sketches of
the soil reactions.
unavoidable uncertainty on the exact location of the emergence
The following trends are noted: of any fault, this means avoiding to build “in close proximity”
• On dense sand, for S/B = 0.2 the increase of po leads to an to the fault
increase of the effective foundation width b from 0.60B to
0.75B, reducing the maximum (unsupported) cantilevered 6.2.3 Pile and Caisson Foundations
spans from u1 = 0.25B to 0.20B (on the left side) and from The role of piles in supporting structures straddling seismic
u2 = 0.2B to 0.10B (on the right side). For S/B = 0.5 there faults is far from clear. Circumstantial evidence from recent
develop two unsupported spans: one (u2) cantilevered on the earthquakes has implicated the piles in some structural damage
right side (towards the hanging wall) and the other (u1) — see for example the analysis of the damage of the pile-
doubly-supported under the left half of the foundation; thus supported Attaturk Stadium in Denizevler during the Kocaeli
there exist two areas of contact: a small (b1) and a larger Earthquake (Anastasopoulos & Gazetas 2007). Systems “tied”
(b2). The increase of po leads to increasing b2 and to the different blocks of the fault may indeed be vulnerable.
decreasing u1 and u2 , while b1 remains essentially the same. Reference is made to Gazetas et al (2007a,b) for a more detailed
exposition of the behaviour of piled foundations under a normal
In stark contrast, in the case of S/B = 0.8 the increase of po
fault rupture.
does not seem to play a significant role.
Rigid caisson foundations are clearly advantageous. The
• On loose sand, the effective width is invariably much larger faulting-induced deformation causes a more-or-less rigid-body
than on dense sand. The most noteworthy effect of displacement and rotation that are in general smaller than those
increasing po is in the case of S/B = 0.5: the unsupported of surface or piled foundations. Reference is made to
spans disappear and full contact is established for po > Anastasopoulos et al (2008a) for more information on the
60kPa. But the biggest beneficial effect of increasing po is subject and an application to an actual bridge problem in
the reduction of the rotation experienced by the foundation. Greece.
The engineer could use Fig. 6.6 to preliminarily design the
foundation raft. But this should be done only if he cannot
reliably avoid building “on the fault”. In view however of the
2922 B. Simpson et al. / State of the art Report: Analysis and Design

6.3 Slender structures on shallow foundations: mobilisation
of bearing capacity mechanisms
6.3.1 Conventional wisdom and the need for change
Seismic design of structures recognises that highly inelastic
material response is unavoidable under the strongest possible
shaking. Displacements as large as 3 times the yield
displacement (in earthquake terminology “ductility” of 3) or
more are usually allowed to develop under seismic loading. This
implies that the strength of a number of critical bearing
elements is fully mobilised. In the prevailing structural
terminology “plastic hinging” is allowed as long as the overall
structural stability is maintained.
By contrast, a crucial goal of current practice in seismic
“foundation” design, particularly as entrenched in the respective
codes (e.g. EC8), is to avoid the mobilisation of “strength” in
the foundation. In structural terminology: no “plastic hinging”
is allowed in the foundation, below the ground surface.
Thus the conventional approach to seismic foundation design
introduces factors of safety against sliding and against
exceedance of ultimate capacity, in a way similar to the
traditional static design. This approach involves two
consecutive steps of structural and foundation analysis:
a. Dynamic analysis of the structure is first performed, in
which the soil is modelled as an elastic medium represented
by suitable translational and rotational springs (and,
sometimes, with the associated dashpots). The dynamic
forces and moments transmitted onto the foundation are
derived from the results of such analyses, after the column
forces have been reduced by dividing with a ductility-
capacity dependent factor. (In EC8 this factor is called
“behaviour” or “q” factor. It varies between 1 and 4;
depending on mainly the ability of the superstructure to
undergo safely large inelastic deformations.)
b. The foundations are then designed in such away that these
transmitted horizontal forces and overturning moments,
increased by “overstrength” factors, would not induce
sliding or bearing capacity failure.
The use of “overstrength” factors is necessitated by the so-
called “capacity design” principle, under which plastic hinging
is allowed only in the structural elements — not in the
below−ground (and hence uninspectable) foundation and soil.
Therefore, structural yielding of the footing and mobilisation of
bearing capacity mechanisms is not allowed. However, there is
a growing awareness in the profession of the need to consider
soil-foundation inelasticity, in analysis and perhaps even in
design (see: Pecker (1998), Paolucci (1997), Martin & Lam
(2000), Allotey & Naggar (2003)). This need has emerged
from:
• The very large accelerations and velocities recorded in
earthquakes, which would impose enormous inelastic
demands to structures if soil−foundation “yielding” would
not effectively limit the induced accelerations.
• Seismic retrofitting of a structure increases the shear
capacity of some elements and hence the forces onto their
foundations; it might not be feasible to undertake them
elastically. A (stiff) concrete shear wall inserted to upgrade
a frame carries most of the inertia-driven shear, and thereby
transmits a disproportionately large horizontal force and
overturning moment onto the foundation. If uplifting,
sliding, and mobilisation of bearing capacity failure
mechanisms are correctly taken into account, the shear wall
“sheds” off some of the load onto the columns; the opposite
is erroneously the case when such inelastic action is
disallowed.
• Many slender historical monuments have apparently
survived several strong seismic motions in their (often long)
life. While under static conditions they would have easily
toppled or otherwise failed, it appears that sliding at, and
especially uplifting from, their base during oscillatory
seismic motion has been a key to their survival. These
phenomena cannot therefore be ignored.
• Compatibility with state of the art structural earthquake
engineering is another reason to compute the complete
inelastic lateral load−displacement or load−rotation
response of the foundation system, to progressively
increasing loads up to collapse. Otherwise the
“performance-based” structural analysis, will be incomplete.
It is therefore logical to extend the inelastic analysis to the
supporting foundation−soil system.
Figure 6.7: Top: problem geometry. Bottom: the two studied
foundation solutions.
6.3.2 Towards a new design philosophy: “plastic hinging” in
soil-foundation systems
Excluding structural yielding in the isolated footing or the
foundation beam, three types of nonlinearity can take place and
modify the overall structure–foundation response:
a. Sliding at the soil−foundation interface. This would happen
whenever the transmitted horizontal force exceeds the
frictional resistance. As pointed out by Newmark (1965),
thanks to the oscillatory nature of earthquake shaking, only
short periods of exceedance usually exist in each direction;
hence, sliding is not associated with failure, but with
permanent irreversible deformations, as will be shown in the
subsequent section of this paper.
b. Separation and uplifting of the foundation from the soil.
This happens when the overturning moment tends to
produce net tensile stresses at the edges of the foundation.
Thanks again to the oscillatory nature of the seismic
shaking, the ensuing rocking oscillations in which uplifting
takes place do not lead to overturning of the structure.
There is not detriment to the vertical load carrying capacity
and the consequences in terms of induced vertical
settlements may be minor. Moreover, in many cases,
footing uplifting is beneficial for the response of the
superstructure, as it helps reduce the ductility demands on
columns.
c. Mobilisation of bearing capacity failure mechanisms in the
supporting soil. Such inelastic action is almost unavoidable
with uplifting of the foundation. In static geotechnical
analysis large factors of safety are introduced to ensure that
bearing capacity modes of failure are not even approached.
In conventional seismic analysis, bearing capacity is
avoided thanks to an “overstrength” factor of about 1.40.
(Note that this factor multiplies the (maximum) design
 B. Simpson et al. / State of the art Report: Analysis and Design 2923

moment at the base of the superstructure; it is this increased
moment that must be carried by the foundation–soil
system.) The oscillatory nature of seismic shaking,
however, allows again the mobilisation (for a short period
of time) of the maximum soil resistance along a
continuous (“failure”) surface. No collapse or
overturning failure occur, as the applied causative
moment quickly reverses, and a similar bearing-capacity
mechanism may develop under the other edge. The
problem again reduces to computing the inelastic
deformations, i.e. permanent rotation.
6.3.3 Results of a comparative study
To illustrate the interaction between soil, foundation, and
structure under strong seismic shaking mobilising inelastic
deformations in the soil we have selected the system
portrayed in Figure 6.7. A single-column concrete bridge
pier, 3m in diameter and about 12m high, carries a deck load
of 12MN (mass of 1200Mg). It is founded with a surface
square foundation (side B) on a 25m thick stiff clay deposit
(Su = 150kPa).

Figure 6.8: Computed displacement response time-histories for the two designs of Figure 6, for the “small” intensity
(Kalamata 1986) motion.

Figure 6.9: Computed moment−rotation and settlement−rotation hysteretic response of the two foundations of Figure 6.6,
for the “small” intensity (Kalamata 1986) motion.
2924 B. Simpson et al. / State of the art Report: Analysis and Design

Two foundation solutions are examined:
a. B = 11m. This derives from a conventional slightly-
conservative design, which leads to a large static
bearing−capacity factor of safety (FSV ≈ 5.6) and a
pseudo-static bearing-capacity factor FSE ≈ 2, for a code-
specified design spectrum having A = 0.24g (horizontal),
corresponding to soil category “B”, and an estimated
“behaviour” or “q” factor of 2. Although the value of 2
is high for a seismic FS, it was chosen in the anticipation
of much-stronger acceleration histories than those
specified by the A = 0.24g code spectrum. It is thus quite
possible that during a design or stronger–than–design
event structural (bending) plastic hinging will develop at
the base of the column, with a rather minimal inelastic
action in the soil or the soil–footing interface ⎯ a
typically prudent conventional design.
b. B = 7 m. This corresponds to the new design concept,
where significant plastic deformation is allowed to take
place in the foundation–soil system, to the point of
mobilisation of the bearing–capacity failure mechanisms.
These may develop alternatingly on either side under the
footing, as large cyclic overturning moments arise during
shaking. This design is barely adequate under static
vertical loads (FSV ≈ 2.8); under the design earthquake it
leads to a pseudo-static FSE = 0.50 ⎯ well below unity to
be acceptable within conventional engineering thinking.
It is therefore expected that during shaking by a
design−level, and especially by an above−design−level,
ground motion the soil “failure” mechanisms will
develop. The question is what the consequences will be
for the foundation and the superstructure, and how the
computed responses of the two systems of Fig. 6.6 differ
from one-another.

Figure 6.10: Computed displacement response time-histories for the two designs of Figure 6, for the “high”
intensity (Takatori 1995) motion.

Figure 6.11: Computed moment−rotation and settlement−rotation hysteretic response of the two
foundations of Figure 6.6, for the “high” intensity (Takatori 1995) motion.
 B. Simpson et al. / State of the art Report: Analysis and Design
This is demonstrated with Figures 6.8 - 6.11. Two
different real accelerograms are used as excitation:
• The Kalamata Administration Building record of the 1986
Ms = 6.2 earthquake (Gazetas et al 1991). With an A ≈
PGA ≈ 0.26g but a response spectrum with values smaller
than those of the design code spectrum at the period range
of interest, this motion will be referred to as “Small”
Intensity excitation.
• The Takatori record of the 1995 MJMA = 7.2 Kobe
Earthquake, which in addition to its high PGA, 0.63 g,
has spectral values in excess of 1.5g over a very wide
period range (0.30 sec – 1.20 sec) . It thus undoubtedly
constitutes a “High” Intensity excitation ⎯ substantially
larger than a design excitation.
Figures 6.8 and 6.9 show the results for the “Small”
Intensity shaking. Specifically, Figure 6.8 plots the time
histories of the main components of superstructure
displacement:
• the displacement of superstructure mass due to foundation
rocking: uș (in blue)
• the (additional) displacement of the superstructure,
representing distortion of the column due to bending: ustr
(grey)
• the algebraic sum of the above two: utot = ustr + uș
(black)
The lateral displacement of the foundation is not shown, as
it is quite secondary for the particular chosen slender system.
Figure 6.9 refers to the response of the foundation. It plots
the dynamic overturning moment-rotation, M − ș, relation
and the dynamic settlement–rotation, w−ș, relation for each
of the two foundations. The conclusions emerging from the
figures are clear:
• The conventional foundation (B = 11 m) experiences very
small nearly-elastic rotation, with a small accumulation of
cyclic settlement (≈ 2.5 cm). By contrast, the column
experiences large distortion, almost 10 cm, as if the
structure were responding on a fixed base.
• The “daring” foundation design, B = 7 m, experiences
very large rotation, with uș reaching 12 cm. The
significant inelastic action in the soil is reflected in the
highly-hysteretic M − ș relationship, as well as in an
appreciable accumulated foundation settlement ( ≈ 6 cm).
By refreshing contrast, the structural distortion has been
limited to merely 2 cm ⎯ indicative of almost elastic
column response.
In similar fashion, Figures 6.10–6.11 compare the results
for the “High” Intensity (Takatori) excitation. The following
is a summary of important conclusions from these plots:
• The conventional foundation design, B = 11 m, with too
little help from the (“unyielding”) foundation, cannot
cope with the huge accelerations of the Takatori record.
mechanisms. The maximum rotation șmax reaches 0.036
rad, corresponding to a substantial deck displacement uș
≈ 50 cm. Accumulated settlement: wmax = wres ≈ 26
cm. The superstructure, however, hardly deforms and
thus remains safe despite the much-larger-than-design
ground shaking. With a warning: the above significant
displacements (26 cm and 50 cm) may imply such large
differential settlements between adjacent foundations and
differential displacements between adjacent piers that
indirect structural damage is unavoidable. Note also that
the developing accelerations in the structure are also
smaller when the soil is yielding and the foundation
uplifts ⎯ an additional benefit effect from foundation
plastic “hinging”.
One might arguably consider the above footing−related
deformations as excessive. Notice, however, that these were
peak values; the residual rotation and displacement appear to
2925
be very small ⎯ for this particular excitation, undoubtedly
coincidentally, they almost vanished!
In conclusion, inelastic “failure” mechanisms could be
allowed to develop in soil-foundation systems designed for
strong seismic excitation. Mobilisation of such mechanisms
does not usually lead to failure; it may in fact prove quite a
beneficial way to save the structure.
Two supporting case histories are mentioned here: (a) the
settlement of slender buildings in Adapazari during the 1999
Kocaeli Earthquake, despite mobilisation of bearing-capacity
“failure” mechanisms (as tentatively explained by Gazetas et
al 2003). (b) the behaviour of the Kobe harbour breakwaters
with no overturning and only small permanent lateral
displacements and rotations; whereas by contrast, the uni-
directionally moving soil-supporting quay walls (caissons
identical to the breakwaters) suffered huge seaward
displacements and rotations (Sekiguchi et al 1996, Dakoulas
and Gazetas 2007).
7 CONCLUDING REMARKS
This paper has presented highlights from recent developments
in some important areas of geotechnical analysis and design.
Whilst analysis plays an important role, the process of design
is wider, incorporating empirical methods based on carefully
recorded experience, together with adequate factors or
margins to provide society with the level of safety and
serviceability recognised as needed on the basis of experience
and tradition. Provision of adequate contractual processes are
also an integral part of design.
In the development of codes and standards, attempts are
being made to rationalise the safety processes, though
reference to past practice is still deemed to be irreplaceable.
Similarly, the valuable results of advanced analytical
techniques must be reviewed and interpreted in the light of a
thorough understanding of the real behaviour of structures
interacting with the ground.
The authors hope that pooling experience from different
continents, varying geologies, onshore and offshore will
enable geotechnical engineers to provide more reliable
analysis and to derive sound designs.
ACKNOWLEDGEMENTS
Dr Simpson and Dr Morrison gratefully acknowledge the
technical and financial support of Arup Geotechnics, in
particular the help of Jon Shillibeer and Kelly Byrne.
Valuable advice has been provided by Professor Roger Green,
Dr Dennis Becker of Golder Associates, Professor KK Phoon
of National University of Singapore and Dr Masahiro Shirato,
CAESAR, PWRI, Japan
The work reported by Professor Gazetas was conducted
for the “DARE” project, financially supported by a European
Research Council (ERC) Advanced Grant under the “Ideas”
Programme in Support of Frontier Research (Grant
Agreement 228254).
REFERENCES
AASHTO. 2008. LRFD Bridge Design Specifications, 4th edition,
2007 with 2008 Interim Revisions. American Association of
State Highway and Transportation Officials.
ACI 224R-01. 2008. American Concrete Institute 224R – Control of
Cracking in Concrete Structures.
ACI 318. 2008. American Concrete Institute 318 - Building Code
Requirements for Structural Concrete.
American Association of State Highway and Transportation Officials
(AASHTO). 2002. Standard Specifications for Highway Bridges.
2926 B. Simpson et al. / State of the art Report: Analysis and Design
American Railway Engineering and Maintenance-of-Way
Association (AREMA). 2008. Manual for Railway Engineering.
Anastasopoulos, I. Bransby, F. 2008b. Design of Slab Foundations
against Seismic Faulting. Part 2: Soil – Structure Interaction.
Proc. 3rd National Conference of Earthquake Engineering and
Engineering Seismology, Athens, 5–7 November, Paper 1958.
Anastasopoulos, I. Gazetas, G. Drosos, V. Georgarakos, T. &
Kourkoulis, R.. 2008a. Design of bridges against large tectonic
deformation. Earthquake Engineeing and Engineering Vibration,
Vol. 7, pp. 345–368.
Anastasopoulos, I. Gerolymos, N. Gazetas, G. & Bransby, M. F.
2008b. Simplified approach for design of raft foundations
against fault rupture. Part I: Free-field. Earthquake Engineering
and Engineering Vibration, Vol. 7, pp. 147–163.
Anastasopoulos, I. Callerio, A. Bransby, M.F. Davies, M.C.R., El.
Nahas, A. Faccioli, E. Gazetas, G. Masella, A. Paolucci, R.
Pecker, A. & Rossignol, E. 2008c. Numerical Analyses of Fault–
Foundation Interaction. Bulletin of Earthquake Engineering,
Special Issue: Integrated approach to fault rupture- and soil-
foundation interaction , Vol. 6, pp. 645–675.
Anastasopoulos, I. Gazetas, G. Bransby, F. Davies, M.C.R. & Nahas,
E. A. 2009. Normal Fault Rupture Interaction with Strip
Foundations. Journal of Geotechnical and Geoenvironmental
Engineering, American Society of Civil Engineers , Vol. 135, No
3, pp. 359-370.
Anastasopoulos, ǿ. 2005. Fault Rupture–Soil–Foundation–Structure
Interaction, Ph.D. Dissertation, School of Civil Engineering,
National Technical University. Athens, pp.570.
API RP 2A-WSD. 2000. Recommended practice for planning,
designing, and constructing fixed offshore platforms - Working
stress design. - 21st edition.
Apostolou, M. Gazetas, G. & Garini, E. 2007. Seismic response of
slender rigid structures with foundation uplifting. Soil Dynamics
and Earthquake Engineering (in press).
Asian Technical Committee No.3 (ATC3). 1997. Manual for
zonation on areas susceptible to rain-induced slope failure, 81p.
Barnes, I. & Sear, L. 2004. Ash Utilisation from Coal-Based Power
Plants. Report No. COAL R274, DTI/Pub URN 04/1915. UK
Department of Trade and Industry.
Becker, D.E. 1996. Eighteenth Canadian Geotechnical Colloquium:
Limit states design of foundations. I: An overview of the
foundation design process. Canadian Geotech J, 33(6), 956-983.
Becker, D.E. 2008. Personal communication.
Berezantzev, V.C. Khristoforov, V. & Golubkov, V. 1961. Load
bearing capacity and deformation of piled foundations. Proc 5th
Int. Conf. Soil Mech. Found. Engng., Paris., 11-15.
Berill, J.B. 1983. Two-dimensional analysis of the effect of fault
rupture on buildings with shallow foundations. Soil Dynamics
and Earthquake Engineering, Vol. 2, No. 3, pp. 156-160.
Bilfinger, W. 2005. Design Procedure for Groundwater Loads on
Tunnel Linings., Symposium on Waterproofing of Underground
Structures, San Paulo Brazil.
Bolton, M.D. 1987. The strength and dilatancy of sands.
Géotechnique, 37, No. 2, 219-226.
Bolton, M.D. 1993. What are partial factors for? In Proceedings of
the International Symposium on Limit State Design in
Geotechnical Engineering, Copenhagen, May 26-28, 1993.
Sponsored by the Danish Geotechnical Society. Vol. 3, pp. 565-
583.
Bond, A. & Harris, A. 2008. Decoding Eurocode 7. Taylor and
Francis, London.
Bond, A.J. & Simpson, B. 2009. Pile design to Eurocode 7 and the
UK National Annex. Submitted to Geotechnical Engineering.
Thomas Telford.
Boverket. 1995. Design Regulations BKR 94 – Mandatory provisions
and general recommendations. Swedish Board of Housing,
Building and Planning.
Brandl, H. 2006. Energy Foundations and other thermo-active
ground structures. Géotechnique 56, No 2, 81-110.
Bray, J.D. 1990.). The effects of tectonic movements on stresses and
deformations in earth embankments. Ph.D. Dissertation,
University of California, Berkeley.
Bray, J.D., Ashmawy, A., Mukhopadhyay, G., Gath E.M. 1993. Use
of Geosynthetics to Mitigate Earthquake Fault Rupture
Propagation Through Compacted Fill. Proceedings of the
Geosynthetics ’93 Conference, Vol. 1, pp. 379-392.
Bray, J.D., Seed, R.B., & Seed, H.B. 1994b. Analysis of Earthquake
Fault Rupture Propagation through Cohesive Soil. Journal of
Geotechnical Engineering, ASCE, Vol. 120, No.3, March, pp.
562-580.
Brinch Hansen, J. 1961. The ultimate resistance of rigid piles against
transversal forces. Danish Geotechnical Institute Bulletin No.
12.
BS5400. 1988. Steel, concrete and composite bridges. BSI, London.
BS8002. 1994. British Standards Institution Code of Practice for
Earth Retaining Structures.
BS8110. 1997. Structural use of concrete. BSI, London.
BSI. 2004. BS EN 1997:1: 2004. Eurocode 7: Geotechnical design -
Part 1: General rules.
BSI. 2007. UK National Annex to Eurocode 7: Geotechnical design -
Part 1: General rules.
BSI. 2007a. BS EN 1997–2:2007 Eurocode 7 Geotechnical design,
Part 2: Ground Investigation and testing.
BSI. 2007b. UK National Annex to Eurocode 7: Geotechnical design
- Part 1: General rules.
Burland, J.B, Potts, D.M. &Walsh, N.M. 1981. The overall stability
of free and propped embedded cantilever walls. Ground
Engineering 14(5), 28-38.
Butcher AP, Powell, J.J.M & Skinner, H.D. 2006a. Reuse of
Foundations for Urban sites: A best practice handbook.
IHS/BRE press.
Butcher AP, Powell, J.J.M & Skinner, H.D (editors), 2006b.
Proceedings of International Conference on Reuse of
Foundations for Urban sites. IHS/BRE press.
Cathart, A. 2003. Channel Tunnel Rail Link: A Contract Partnership.
Civil Engineering Vol. 156, Issue 5, pp 41-44.
Central Electricity Generating Board. 1965. Report of the Committee
of Inquiry into Collapse of Cooling Towers at Ferrybridge
Monday 1 November 1965. Central Electricity Generating
Board, London.
Chapman, T. Connolly, M. Nicholson, D. Raison, C & Yeow. H.
1999. Advances in understanding of base grouted pile
performance in very dense sand. International symposium and
exhibition in tunneling construction and piling, London, 1999, pp
57-69.
Chau, C. Nicholson, D. & Soga, K. 2006. Comparison of embodied
energy of four different retaining wall systems. Proc. Int. Conf.
Reuse Found. for Urban Sites (RuFus 2006), 277–285.
Cheng, Y.M. 2004. Géotechnique 54, No 2, 149-150.
Clausen, C.J.F. Aas, P.M & Karlsrund., K. 2005. Bearing capacity of
driven piles in sand, the NGI approach. Int. Symp. on Frontiers
in Offshore Geotechnics, Perth, Australia (ISFOG 2005).
Cremer, C., Pecker, A. & Davenne, L. 2002. Modelling of nonlinear
dynamic behaviour of a shallow strip foundation with macro-
element. Journal of Earthquake Engineering 6, No. 2, 175-211.
Dahlberg,R and Ronold, KO (1993) Limit state design of offshore
foundations. Proc Int Symp Limit state design in geotechnical
engineering, Vol 2, pp491-500. Danish Geotechnical Society.
Dakoulas, P. & Gazetas, G. 2008. Insight to Seismic Earth and Water
Pressures Against Caisson Quay Walls. Geotechnique, Vol. 58,
No. 2, pp 95-113.
Dansk Ingeniorforening. 1984. Danish Code of Practice for
Foundations Engineering (Dansk Standard DS 415). English
translation.
Det Norske Veritas. 1993. Rules for classification of fixed offshore
installations.
Duddeck, H. & Erdmann, J. 1982. Structural design models for
tunnels. Tunnelling 82, pp 83-91.
EN1990. 2002. Eurocode: Basis of design. BSI, London. (BS EN
1990:2002).
EN1992. 2004. Eurocode 2 – Design of concrete structures. BSI,
London. (BS EN 1992:2004).
EN1997. 2004. Eurocode7: Geotechnical design. BSI, London. (BS
EN 1997:2004).
EN1997-1. 2004. Eurocode7: Geotechnical design - Part 1: General
rules. BSI, London. (“EC7”).
Eurocode EC8. 2000. Structures in Seismic Regions. Part 5:
Foundations, retaining structures, Geotechnical Aspects.
Faccioli, E. Anastasopoulos, I. Callerio, A. & Gazetas, G. 2008. Case
histories of fault–foundation interaction. Bulletin of Earthquake
Engineering, Special Issue: Integrated approach to fault rupture-
and soil-foundation interaction, Vol. 6, pp. 557–583.
 B. Simpson et al. / State of the art Report: Analysis and Design
FEMA 356. 2000: Prestandard and Commentary for the Seismic
Rehabilitation of Buildings.
Fleming, W.G.K, Weltman, A.J, Randolph, M.F & Elson, W.K.
1992. Piling engineering. end edn. WIley.
Fleming, W.G.K, Weltman, A.J, Randolph, M.F & Elson, W.K.
2008. Piling engineering. 3rd edition. Taylor and Francis Ltd.
Foye, K.C, Salgada, R. & Scott, B. 2006a. Assessment of variable
uncertainties for reliability-based design of foundations. J
Geotech Geoenviron Eng, 132(9), 1197-1207.
Foye, K.C. Salgado, R. & Scott, B. 2006b. Resistance factors for use
in shallow foundation LRFD. J. Geotech. Geoenviron. Eng.,
132(9), 1208–1218.
Fugro. 2004. Fugro Engineers B.V., Axial pile capacity design
method for offshore driven piles in sand.
Gaba, A.R, Simpson, B. Powrie, W. & Beadman, D.R. 2003.
Embedded retaining walls: guidance for economic design.
CIRIA Report C580.
Gajan, S. Kutter, B. Phalen, J. Hutchinson, T. & Martin, G. 2005.
Centrifuge modelling of load-deformation behaviour of rocking
shallow foundations. Soil Dynamics and Earthquake
Engineering, (in press).
Gazetas, G. Anastasopoulos, I. Apostolou, M. 2007. Shallow and
Deep Foundations Under Fault Rupture or Strong Seismic
Shaking. Geotechnical Earthquake Engineering, Springer,
Chapter 9, pp. 185-217.
Gazetas, G. Dakoulas, P. & Papageorgiou, A. 1990. Local-Soil and
Source Mechanism Effects in the 1986 Kalamata
Earthquake. Earthquake Engineering & Structural Dynamics,
Vol. 19, pp. 431-456.
Gazetas, G. Pecker, A. Faccioli, E. Paolucci, R. & Anastasopoulos, I.
2008. Design Recommendations for Fault–Foundation
Interaction. Bulletin of Earthquake Engineering, Special Issue:
Integrated approach to fault rupture- and soil-foundation
interaction, Vol. 6, 677–687.
Gazetas, G. & Apostolou, M. 2004. Nonlinear soil-structure
interaction: foundation uplift and soil yielding. Proc. 3rd US-
Japan Workshop on Soil-Structure Interaction, USGS, M.
Todorovska and M. Celebi editors, Menlo Park, CA.
GEO. 2006. Foundations design and Construction. Geo Publication
1/2006, The Government of Hong Kong.
Green, R. 2008. Personal communication.
Honjo, Y.Y. Kikuchi, M. Suzuki, K. Tani & M. Shirato. 2005. JGS
Comprehensive Foundation Design Code: Geo-code 21, Proc.
16th ICSMGE, pp.2813-2816, Osaka.
Housner, G. 1963. The Behavior of Inverted Pendulum Structures
During Earthquakes. Bulletin of the Seismological Society of
America. 53(2), 404-417.
Hovarth, R.G. Kenny, T.C. & Trow, W.A. 1980. Results of tests to
determine shaft resistance of rock socketed drilled piers. Int.
conf. on structural foundations on rock, Sydney.
Iai, S. Matsunaga, Y. & Kameoka, T. 1992. “Strain space plasticity
model for cyclic mobility,” Soils and Foundations, Vol.32, No.2,
pp.1-15, Ishihara, K. and Yasuda, S. 1975. Undrained deformation
and liquefaction of sand under cyclic stresses, Soils and Foundations,
Vol.23, No.6, pp.29-35.
Iwasaki, T. Tatsuoka, F. Tokida, K. & Yasuda, S. 1978. A prediction
method for assessing soil liquefaction potential based on case
studies at various sites in Japan, Proc. of the 2nd Int. Conf. on
Microzonation, Vol.2, pp.885-896.
Japan Road Association. 1986. Earthwork manual for soft ground.
(in Japanese).
Japanese Geotechnical Society. 2006. Japanese Geocode 21:
Principles for Foundation Designs Grounded on a Performance-
based Design Concept, JGS 4001-2004.
Japanese Geotechnical Society. 1998. Remedial measures against soil
liquefaction, 443p.Balkema.
Japanese Geotechnical Society. 2004. Reconnaissance report on the
2003 Tokachi-oki earthquake. (in Japanese).
Japanese River Association. 1997. “Manual criteria for river works,”
(in Japanese).
Japanese Society of Soil Mechanics and Foundation Engineering.
1994. Reconnaissance report on the 1993 Kushiro-oki
earthquake and the 1993 Notohaoto-oki earthquake. (in
Japanese).
Jardine, R. Chow, F. Overy, R. & Standing, J. 2005. ICP Design
Methods for Driven Piles in Sands and Clays. Thomas Telford,
London.
2927
Johansson, J. & Konagai, K. 2004. Fault induced permanent ground
deformations–Simulations and experimental verification.
Proceedings of the 13th World Conference on Earthquake
Engineering, August 1–6, Vancouver, Canada.
Johnston, I.W. & Lam T.S.K. 1989. Shear behaviour of regular
triangular concrete/rock joints – analysis. Journal of
Geotechnical Engineering, ASCE, Vol 115.5, pp 711-725.
Kanto Branch of the Japanese Geotechnical Society. 2007. Guideline
for site investigation, analyses and countermeasures of existing
reclaimed housing lots against earthquakes, 201p. (in Japanese)
Kawashima, K. & Hosoiri, K. 2005. Rocking Isolation of Bridge
Columns on Direct Foundations. Journal of Earthquake
Engineering, 6(2).
Klotz, E.U. & Coop, M.R. 2001. An investigation of the effect of soil
state on capacity of driven piles in sands. Géotechnique 51, No.
9, 733-751.
Lambe, T.W & Whitman, R.V. 1979. Soil mechanics, SI version.
Wiley.
Lazarte, C.A. & Bray, J.D. 1995. Observed Surface Breakage due to
Strike-Slip Faulting, Third International Conference on Recent
Advances in Geotechnical Engineering and Soil Dynamics, Vol.
2, pp. 635–640.
Lehane, B.M. & Randolph, M.F. 2002. Evaluation of a minimum
base resistance for driven piles in siliceous sand. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, Vol.
130.6, pp 656-658.
Lehane, B.M. Schneider, J.A. & Xu, X. 2005a. CPT based Design of
Driven Piles in Sand for offshore structures. University of
Western Australia report Geo 05345.
Lehane, B.M. Schneider, J.A. & Xu, X. 2005b. The UXA-05 method
for prediction of axial capacity of driven piles in sand. Frontiers
in Offshore Geotechnics: ISFOG 2005, Taylor and Francis
Group, pp 683-689.
Lord, J.A, Clayton, C.R.I. & Mortimore, R.N. 2002. Engineering in
Chalk. Ciria Report No C574.
Lyamin, A. V. Salgado, R. Sloan, S. W. & Prezzi, M. 2006. Three
dimensional bearing capacity of footings in sand. Working
paper, Purdue Univ.
//http://www.ecn.purdue.edu/~rodrigo/papers/.
Magnus, R. Teh, C.I. & Lau, J.M. 2005. Report on the Incident at the
MRT Circle Line worksite that led to the collapse of the Nicoll
Highway on 20 April 2004. Subordinate Courts, Singapore.
Makris, N. & Roussos, Y.S. 2000. Rocking Response of Rigid Blocks
under Near-Source Ground Motions. Geotechnique, 50(3), 243-
262.
Mandolini, A. Russo, G. & Viggiani, C. 2005. Pile foundations:
experimental investigation, analysis and design. State of the Art
Report. Proc. 16th International Conference on Soil Mechanics
and Geotechnical Engineering, Osaka. Vol. 1, 177-213.
Martin, C.M. & Randolph, M. F. 2006. Upper-bound analysis of
lateral pile capacity in cohesive soil. Géotechnique, Vol. 56.2,
pp 141-145.
Matsuo, M. and Kawamura, K. 1977. Diagram for construction
control of embankment on soft ground. Soils and Foundations,
Vol.17, No3, pp37-52.
Mayne, P.W. & Kulhawy, F.H. 1982. Ko – OCR relationship in soil.
Geotechnical Engineering, ASCE. Vol 108.6; 851-872
Mehta, P.K. 2001. Reducing the Environmental Impact of Concrete.
Concrete International, October 2001.
Mehta, P.K. 2004. High Performance, High Volume Fly Ash
Concrete For Sustainable Development Proc Int Workshop on
Sustainable Development and Concrete Technology, Beijing
Iowa State University.
http://www.ctre.iastate.edu/pubs/sustainable/index.htm.
Meyer, C 2005. Concrete as a Green Building Material. Invited
Lecture, Proceedings of the Third Int. Conference on
Construction Materials, ConMat'05, Vancouver, Aug. 22-25.
Monzon, J.C. 2006. Review of CPT based design methods for
estimating axial capacity of driven piles in siliceous sand.
M.Eng thesis submitted to MIT, USA.
Nakai, T. Muir Wood, D. & Stone, K.J.L. 1995. Numerical
Calculations of Soil Response Over a Displacing Basement. Soils
and Foundations, Vol. 35, No.2, pp. 25–35.
Nakamura, Y. & Murakami, Y. 1980. A study on the seismic stability
of embankments of three big rivers in Nobi Plain, 34th Technical
Conference of Ministry of Construction, pp.96-104. (in
Japanese).
2928 B. Simpson et al. / State of the art Report: Analysis and Design
Newmark, N. M. 1959. A method of computation for structural
dynamics, ASCE, Vol.85, EM3, pp.67-94.
Nicholson, D Chapman, T & Morrison, P. 2002. Pressuremeter
proves its worth in London’s Docklands. Ground Engineering
March 2002, pp 32-34.
Nicholson, D.P. Dew, C.E. & Grose, W.J. 2006. A Systematic ‘Best
Way Out’ Approach using Back Analysis and the Principles of
the Observational Method. International Conf on Deep
Excavations, 28-30 June 2006, Singapore.
Oka, F. Yashima, A. Tateishi, A. Taguchi, Y. & Yamashita, S. 1999.
“A cyclic elasto-plastic constitutive model for sand considering a
plastic-strain dependence of the shear modulus.” Geotechnique
Vol.49, No. 5, pp.661-680.
Orr, T.L.L. 2005. Model Solutions for Eurocode 7 Workshop
Examples. Trinity College, Dublin.
Osman A.S. and Bolton M.D. 2006. Design of braced excavations to
limit ground movements. Proceedings of Institution of Civil
Engineers, Geotechnical Engineering 159 (3), 167-175.
Osman, A.S. Yeow, H.C. Bolton, M.D. 2004. Estimation of
Undrained Settlement of Shallow Foundations on London Clay. -
First International Conference on Structural and Foundation
Failures August 2-47-9, 2004, Taipei, Taiwan Singapore.
Padfield, C.J. & Mair, R.J. 1984. Design of retaining walls embedded
in stiff clay. CIRIA Report 104.
Pecker A. 2005. Design and Construction of the foundations of the
Rion-Antirrion Bridge. Proceedings, 1st Greece Japan Workshop
on Seismic Design, Observation and Retrofit of Foundations
Pells, P.J.N., Rowe, R.K., & Turner, R.M. 1980. An experimental
investigation into side shear for socketed piles in sandstone. In
Proceedings of the International Conference on Structural
Foundations on Rock, Sydney, pp. 291–302.
Pietruszezak, S.T., & Mroz, Z. 1981. Finite element analysis of
deformation of strain softening materials. International Journal
for Numerical Methods in Engineering, Vol. 17, pp. 327–334.
Pollalis, S.N. 2006. Channel Tunnel Rail Link, Risk Transfer and
Innovation in Project Delivery., Harvard Design School,
November 2006.
Potts, D. M. Dounias, G. T. & Vaughan, P. R. V. 1987. Finite
element analysis of the direct shear box test. Géotechnique, Vol.
37, No. 1, pp. 11–23.
Poulos, H. G. & Davids, J. 2005.Foundation design for the Emirates
Twin Towers, Dubai. Canadian Geotechnical Journal, Vol 42, pp
716-730.
Poulos, H.G. & Davis E. 1980. Pile Foundation Analysis and
Design.
Pristley, N. Seible, F. & Calvi, G.M. 1996. Seismic Design and
Retrofit of Bridges. John Wiley and Sons, Inc., New York.
Public Works Research Institute. 1997. Design and construction
manual for remediation of river dikes against liquefaction,
Research report of PWRI, Vol.3513. (in Japanese).
Railway Technical Institute. 1999. Seismic design code for Japanese
railway facilities, (in Japanese).
Randolph, M.F. 2003a. Science and empiricism in pile foundation
design. Géotechnique Vol. 53.10 847-875.
Randolph, M. F. 2003b. PIGLET, Analysis and design of Pile
Groups, Version 4.2. (User manual).
Randolph, M. Cassidy, M. Gourvenec, S. & Erbich, C. 2005.
Challenges of Offshore Geotechnical Engineering. State of the
Art Report. Proc. 16th International Conference on Soil
Mechanics and Geotechnical Engineering, Osaka. Vol. 1, 123-
176.
Reece, L.C. 1997. Analysis of laterally loaded piles in weak rock.
Journal of Geotechnical and Geoenvironmental Engineering,
Vol. 123.11, pp 1010-1017.
River Bureau. 1995. Technical manual for seismic diagnosis of
existing river dikes, River Bureau, MLIT. (in Japanese)
River Bureau. 2007. Performance-based seismic design criteria for
river facilities (Draft). (in Japanese).
Rock Engineering. 1994. Engineering and Design – Rock
Engineering. US Army Corps of Engineers, Report No EM
1110-1-2908 (http://140.194.76.129/publications/eng-
manuals/em1110-1-2908/toc.htm).
Rostam, S. & Høj, N.P. 2004. Integrated Tunnel Design from Cradle
to Grave, Enhancing Structural Performance and Owner
Confidence. Safe & Reliable Tunnels. Innovative European
Achievements, First International Symposium, Prague 2004.
Roth, W.H. Scott, R.F. & Austin, I. 1981. Centrifuge modelling of
fault propagation through alluvial soils. Geophysical Research
Letters, Vol. 8, No. 6, pp. 561–564.
Rowe, R.K. & Armitage, H.H. 1987. A design method for drilled
piers in soft rock. Canadian Geotechnical Journal, Vol 13.3, pp
324-333.
Salençon, J. & Pecker, A. 1995. Ultimate bearing capacity of shallow
foundations under inclined and eccentric loads. Part II: Purely
cohesive soil without tensile strength. Eur. J. Mech. A/Solids 14,
No. 3, 377-396.
Sasaki, Y. Kano, S. & Matsuo, O. 2004. Research and practices on
remedial measures for river dikes against soil liquefaction,
Journal of Japan Association for Earthquake Engineering, Vol.4,
No.3 (Special Issue), pp.312-335.
Sawada, R. Tanamura, S. Nishimura, A. & Koseki, J. 1999.
Simplified method to subsidence of embankment on liquefiable
ground. Proc. 34th Japan National Conference on Geotechnical
Engineering. pp.2091-2092. (in Japanese).
Schneider, H. R. 1997. Definition and determination of characteristic
soil properties. Contribution to Discussion Session 2.3, XIV
ICSMFE, Hamburg. Balkema.
Salgado, R. Lyamin, A. V. Sloan, S. W. & Yu, H. S. 2004. Two- and
three-dimensional bearing capacity of foundations in clay.
Géotechnique 54, No.5, 297-306.
Schulze, H and Duddeck, H 1964. Spannungen in
schildforgetriebenen Tunneln. Beton Stahlbetonb., 59.
Schuppener, B. Simpson, B. Orr, T.L.L. Frank, R. Bond, A.J. 2009.
Loss of static equilibrium of a structure - definition and
verification of limit state EQU. 2nd Int Symp on Geotechnical
Safety and Risk, Gifu.
Seidel, J.P. & Collingwood, B. 2001. A new rock socket roughness
factor for prediction of rock socket shaft resistance. Canadian
Geotechnical Journal, Vol. 38, pp 138-153.
Seifart, R. 2007. Permafrost, A building problem in Alaska.
Published by University of Alaska Fairbanks.
Shirato, M. 2008. Personal communication.
Sidel, J. P. & Haberfield, C.M. 1995a. Towards an understanding of
joint roughness. Int Journal of Rock Mechanics and Rock
Engineering, Vol. 28.2, pp 69-92.
Sidel, J. P. & Haberfield, C.M. 1995b. The axial capacity of pile
sockets in rocks and hard soils. Ground Engineering, Vol. 28.2,
pp 33-38.
Simpson, B. (2007) Approaches to ULS design - The merits of Design
Approach 1 in Eurocode 7. ISGSR2007 First International
Symposium on Geotechnical Safety & Risk pp 527-538.
Shanghai Tongji University, China.
Simpson, B. & Driscoll, R. 1998. Eurocode 7 - a commentary.
Construction Research Communications Ltd, Watford, UK.
Simpson, B. Nicholson, D.P. Banfi, M. Grose, W.G. & Davies, R.V.
2008).Collapse of the Nicoll Highway excavation, Singapore.
Proc Fourth International Forensic Engineering Conference.
Thomas Telford.
Simpson, B. Yeow, H.C. Pillai, A.K. & Grose, W.J. 2006. Benefits
derived from use of 3D finite element analysis in the design of
deep excavations and tunnels. Proc Int Conf on Deep
Excavations 28-30 June 2006, Singapore.
Technical Committee on the 1968 Tokachi-oki earthquake. 1969.
Reconnaissance report on the 1968 Tokachi-oki earthquake. (in
Japanese).
Tohoku Branch, JSCE. 1980. Reconnaissance report on the 1978
Miyagiken-oki earthquake. (in Japanese).
Transportation Research Board. 2007. Guidelines for Using
Environment–Friendly Concrete (EFC) in Transportation
Infrastructure, Sponsoring Committee: AFN10, Basic Research
and Emerging Technologies Related to Concrete.
Tsuchida, T. Sato, T. Hong, Z. Mitsukuri, K. & Sakai, K. 2001. Field
Placing Test of Lightweight Treated Soil at 10m Seawater Depth
in Kumamoto Port. Technical Note of the Port and Airport
Research Institute, no. 1007. http://sciencelinks.jp/j-
east/article/200204/000020020402A0106434.php
Turner, J. 2006. Rock-Socketed shafts for Highway structure
foundations – A Synthersis for Highway Practice. Transport
Research Board, Washington, USA.
US Army. 1994. Engineering and Design - Rock Foundations, US
Army Corps of Engineers, EM 1110-1-2908.
 B. Simpson et al. / State of the art Report: Analysis and Design
Ventouras, K. & Coop, M. R. 2009. On the behaviour of Thanet
Sand: an example of an uncemented natural sand.
Géotechnique, in press.
Vogt, N. & Schuppener, B. 2006. Implementation of Eurocode 7-1
Geotechnical design in Germany. Proc TAIPEI2006 International
Symposium on New Generation Design Codes for Geotechnical
Engineering Practice Nov. 2~3, 2006, Taipei, Taiwan.
Walters, J.V. & Thomas, J.N. 1982. Shear Zone Development In
Granular Materials. Proceedings of the 4th International
Conference on Numerical Methods in Geomechanics, Edomnton,
Canada, May 31 – June 4, Vol. I, pp. 263-274.
White, D.J. & Bolton, M.D. 2005. Comparing CPT and pile base
resistance in sand. Proc. Inst. Civil Eng. Geotechnical
Engineering158, Jan, 2005, 3-14.
Xu, X. & Lehane, B.M. 2005. Evaluation of end-bearing capacity of
closed ended pile in sand from cone penetration data. Proc
ISFOG 2005, Perth, Australia.
Yasuda, S. 2005. Recent several studies and codes on performance-
based design for liquefaction in Japan, Proc. of Geotechnical
Earthquake Engineering Satellite Conference, 16th ICSMGE.,
pp. 46-53.
Yasuda, S. 2007. Countermeasures against seismic failure of existing
reclaimed areas by banking. Keynote Lecture, Proc. of International
Workshop on Earthquake Hazards and Mitigations (EHAM 2007).
Yasuda, S. 2007. Methods for remediation of existing structures
against liquefaction, Proc. of the 4th International Conference on
Earthquake Geotechnical Engineering, Keynote Lecture.
Yasuda, S. Fujioka, K. Shiratori, S. & Mouri, T. 2008. Behaviour and
Design of Expressway Embankments Subjected to Earthquakes,
Proc. of the 3rd Taiwan-Japan Workshop on Geotechnical
Hazards for Large Earthquakes and Heavy Rainfalls, Keynote
Lecture, pp.1-20.
2929
Yasuda, S. Hitomi, T. & Hashimoto, T. 2004. A detailed study on the
liquefaction-induced settlement of timber houses during the 2000
Tottoriken-seibu earthquake, Proc. of the 5th International
Conference on Case Histories in Geotechnical Engineering,
Paper No.3.33.
Yasuda, S. Ideno,T. Sakurai, Y. Yoshida, N. & Kiku, H. 2003.
Analyses for liquefaction-induced settlement of river levees by
ALID. Proc. 12th Asian Regional Conference on Soil Mechanics
and Geotechnical Engineering, pp.347-350.
Yasuda, S. Ideno, T. Sakurai, Y. Yoshida, N. & Kiku, H. 2003.
Analyses for liquefaction-induced settlement of river levees by
ALID, Proc. of the 12th Asian Regional Conference on Soil
Mechanics and Geotechnical Engineering, pp.347-350.
Yasuda, S. Morimoto, I. Kiku, H. & Tanaka, T. 2004.
Reconnaissance report on the damage caused by three Japanese
earthquakes in 2003, Proc. of the 3rd International Conference
on Earthquake Geotechnical Engineering, Vol.1, pp.14-21.
Yazdchi, M. Macklin, S.R. Yeow, H.C. 2005. 3D modelling of
sprayed-concrete-lined tunnels in clay. Journal of Geotechnical
Engineering, 159, 4, 243-250. Thomas Telford.
Yeow, H.C. Nicholson, D. Morrison, P. Coupland, J. & Kassir, M.
2005. Moorhouse, London – Designing for Future Crossrail.
DFI publication Deep Foundations, Winter 2006.
Yeow, H.C. & Prust, R. 2005, "Feasibility of Modelling of Soil-
Structure Interaction Problems using Full Three-dimensional
Finite Element Method", ASCE International Conference on
Computing in Civil Engineering, Mexico.
Yeow, H.C. Wong, C. Pillai, A.K. & Simpson, B. 2005. The Use of a
Substructure Method and Three-dimensional Finite Element
Modelling in Assessment of Damage due to Underground
Tunnelling. 11th International Association of Computer Methods
and Advances in Geomechanics Conference, Italy.